JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY OCTOBER 199 1 VOL. 6 229R ATOMIC SPECTROMETRY UPDATE-INORGANIC MASS SPECTROMETRY AND X-RAY FLUORESCENCE SPECTROMETRY Jeffrey R. Bacon The Macaulay Land Use Research Institute Craigiebuckler Aberdeen A B9 2QJ UK Andrew T. Ellis* Oxford Analytical Instruments Ltd. 19-20 Nuffield Way Abingdon Oxfordshire OX14 1 TX UK John G. Williams NERC ICP-MS Facility Royal Holloway and Bedford New College Department of Geology Egham Hill Egham Surrey TW20 OEX UK Summary of Contents 1 Inorganic Mass Spectrometry 1 .l. Introduction 1.2. Accelerator Mass Spectrometry (AMS) 1.3. Glow Discharge Mass Spectrometry (GDMS) 1.4. Inductively Coupled Plasma Mass Spectrometry (ICP-MS) 1.4.1. General reviews 1.4.2. Fundamental studies 1.4.2.1. Operating parameters and instrument development 1.4.2.2.Calibration 1.4.2.3. Sample introduction 1.4.2.4. Interference effects 1.4.3.1. Geochemistry 1 -4.3.2. Biological and environmental materials 1.4.3.3. Materials 1.4.3. Applications 1.5. Laser Microprobe Mass Spectrometry (LMMS) 1.6. Resonance Ionization Mass Spectrometry (RIMS) 1.7. Secondary Ion Mass Spectrometry (SIMS) 1.8. Spark Source Mass Spectrometry (SSMS) 1.9. Sputtered Neutral Mass Spectrometry (SNMS) 1.1 0. Stable Isotope Ratio Mass Spectrometry (SIRMS) 1.1 1. Thermal Ionization Mass Spectrometry (TIMS) 1.1 2. Other Methods 2 X-ray Fluorescence Spectrometry 2.1. Reviews 2.2. Instrumentation 2.2.1. Excitation 2.2.2. Detectors 2.2.3. Total reflection XRF (TXRF) 2.3.1. Fundamental parameter (FP) calculations 2.3.2.Matrix correction using scattered radiation 2.4.1. Specimen preparation 2.4.2. Thin films 2.4.3. Geological materials 2.4.4. Industrial materials 2.4.5. Environmental analysis 2.4.6. Biological materials 2.4.7. Chemical effects 2.3. Calibration and Data Processing 2.4. Applications The MS section this year continues to provide evidence for the vigorous development of many of the diverse techniques that come under the title of inorganic MS. The plasma techniques (GDMS and ICP-MS) in particular have experienced rapid growth but it is of concern that many reports are abstracts from conferences or meetings rather than full papers. The application of lasers continues to be popular and is an area of considerable growth. Emphasis is placed this year on instrument refinements and technique developments although novel or interesting applications still receive attention.This review year the XRF section places reduced emphasis on applications by concentrating on work which reports particularly interesting or new aspects. Specimen preparation remains an area of particular importance and this is reflected by the degree of coverage in this Update. The instrumentation section has been strengthened and this helped by the continuing level of publication in areas such as TXRF SRXRF and X-ray microfluorescence and new interest in CCD detectors. ~~ * Review Co-ordinator to whom correspondence should be addressed.230R JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY OCTOBER 1991 VOL. 6 1. INORGANIC MASS SPECTROMETRY 1 .l. Introduction The range of instrumental and procedure developments covered in this Update is evidence of the considerable activity in inorganic MS and of its use and acceptance by researchers from a wide range of disciplines.Advances continue to be made in a number of techniques in particular ICP-MS GDMS and SNMS while other estab- lished techniques such as SIMS SIRMS and TIMS are used widely in applications and are often considered the definitive method for a particular measurement. The format of this review remains the same as last year’s in that all areas of atomic MS are covered but topics such as studies of organometallic compounds cluster ions and gaseous metal chemistry are excluded. Emphasis is placed on instrument refinements and technique developments whereas applications are generally excluded unless they are new or unique.This results in less attention being given to the established techniques such as SIMS and TIMS whereas new or developing techniques such as accelerator mass spectrometry (AMS) and GDMS receive greater attention. For typical applications the reader is referred to the Updates on industrial analysis (90/27 19) environmen- tal analysis (90/349 1) and clinical and biological materials (90/3 500). Some overall reviews are worthy of mention. In particular the biennial review of Koppenaal (91/160) of atomic mass spectrometry is highly recommended for it gives in-depth coverage of the whole subject. Background information of value to the reader not totally familiar with a particular technique is given and the comment is informative.Of reduced scope but still of considerable value is the review by Gijbels (90/3590) of elemental analysis of high-purity solids. This gives details of the principles and operational procedures for GDMS LMMS SIMS SNMS and SSMS and makes useful comparisons between the performance and analytical value of each technique. Laser methods are receiving much attention and the lengthy review of Moenke-Blankenburg (9 1 /1250) included mass spectrome- tric techniques amongst the analytical methods con- sidered. Three trends are apparent in instrumental development. The wider availability of cheaper tunable diode lasers sees their ever-increasing application in inorganic MS. The laser microprobe has long been accepted as a powerful technique and laser ablation is also being used as a source of neutral atoms to be post-ionized by either resonant or non-resonant laser radiation. These approaches lead to the second trend in the decoupling of atomization and ionization.Develop- ments of RIMS continue and the power of SNMS is becoming more apparent as more instruments are built to investigate its potential. A number of different sources of atoms are being explored and coupled with a range of ionization methods. The third trend is that towards multi-function instru- ments in particular for surface analysis of high-purity materials. One such development (90/23 77) combines LMMS and SIMS in a time-of-flight (TOF) instrument. Both ion and laser primary beams are available with an electron beam for charge neutralization. Either secondary ions or secondary neutrals following post-ionization by laser beam or electron gun can be analysed.In a further configuration analyte is delivered by a pulsed jet of gas or vapour and subsequently ionized and analysed. A versatile system for quantitative surface analysis and depth profiling includes several surface and auxiliary techniques including SlMS and SNMS (9013438). A commercial high-perform- ance mass spectrometry system designed for isotope ratio and ultra-trace element microanalysis in the earth sciences has the facility to generate ions thermally (TIMS) by ion sputtering (SIMS) or by laser post-ionization (RIMS SNMS) (90/C3566). Interest in ICP-MS by the analytical community contin- ues to grow if the number of abstracts available for review this year is a true reflection.Unfortunately of the approxi- mately 350 abstracts only about 60 were full written reports the remainder being abstracts from conferences or meetings. As reported in last years review few of these pieces of work ever become full papers. What is of greater concern is the number of ‘replicate’ abstracts that appear from different meetings. All have the same lack of data and offer promises of what will be presented. It can only be hoped that more of this potentially interesting work will appear in the scientific literature. The technique continues to gain popularity in Japan although unfortunately many of the available papers are published in Japanese. 1.2. Accelerator Mass Spectrometry (AMS) This technique appears to be entering a new phase in which technique development slows down and the achievements and capabilities are assessed.A survey of operation (91/2293) summarized the experience of the five labora- tories (Oxford Arizona Nagoya Toronto and Gif-sur- Yvette) which have been using Tandetron accelerators for routine analytical work for more than six years. Salient features of each instrument were given and typical applica- tions discussed. It was concluded that the component still requiring most attention is the high-voltage terminal of the accelerator which is influenced by column structure accel- erator tubes and voltage-control electronics. A possible solution to the continued search for ultrasensi- tive techniques in semiconductor surface characterization has been the coupling of a conventional SIMS source to a high-energy tandem accelerator (9 1/2272).This gives highly improved detection limits [ppb (atomic) sensitivity] com- pared with conventional methods and the prospects of single atom detection if nuclear particle counters are used. This extreme sensitivity requires high sputter rates however so the spatial resolution is compromised. A further improve- ment might be achieved if an SNMS source were used to reduce matrix effects and eliminate mass interferences. Another study (91/2294) has found that AMS could be a useful tool for elemental analytical surveys of semiconduc- tor materials. A Cs ion sputtering source was configured with a tandem accelerator to give ultrasensitive detection of impurities. Molecular interferences observed in conven- tional SIMS spectra on for example 31P and 56Fe were removed by dissociation of the silicon-based molecular ions during passage through the stripper canal of the accelerator.Cleanliness of the ion source becomes paramount in such ultrasensitive analyses. Instrumental developments to improve detection limits have allowed the number of radioisotopes studied by AMS to increase and there is a move towards determination of higher mass isotopes. This was illustrated by the novel applications outlined by Rucklidge et al. (9 112292). Deter- mination of 36Cl and 1291 in hydrological studies allowed the monitoring of radioactive waste disposal sites. Submarine hydrothermal systems were studied by measurement of 9Be and l0Be. Low levels of Au and Pt could be measured in individual sulphide and oxide grains in rocks by use of a Cs+ primary ion beam with a diameter of 0.5 mm.Small accelerators (2 MeV Tandetron) were suggested to be suitable only for applications where no isobaric inter- ferences were observed and so were unsuitable for the measurement of 36Cl interfered by 36S but suitable for Au and Pt. The combination of low terminal voltage and high charge state (+ 6) of the detected ions resulted in a low yieldJOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY OCTOBER I99 1 VOL. 6 231R ( ~ 6 % ) but the calibration was linear over four orders of magnitude and the detection limits were t l ng g-1. The sensitivities for Au and Pt differed by a factor of ten however. The application of AMS to radiocarbon analysis has been reviewed (91/2338).The methods used for analysis were described in detail and the problem of conversion from radiocarbon ages to calender ages discussed. The fractiona- tion of the 13C:12C ratio in targets was measured by SIRMS to be (1 k l)o/oo and this was used to correct the l4C:I3C ratios determined by AMS. Application to the measure- ment of ages of bones textiles cultigens and other archaeo- logical artifacts was described. Two interesting applications of 14C dating have been to studies aimed at better understanding the last deglaciation. Hand-picked organic material from Ireland was analysed to establish the date of vegetation development following deglaciation (9 1/235 1). The dates obtained suggested that these events occurred slightly more recently (400 years) than has been traditionally accepted.Benthic foraminifera taken from cores from the Atlantic near Portugal have been dated to calculate a sea-level rise of about 40 m during the millenium which followed 14 500 years BP (91/2352). 1.3. Glow Discharge Mass Spectrometry (GDMS) Advances continue to be made in all aspects of GDMS but especially in two areas the understanding of the fundamen- tal ionization mechanisms and procedures to reduce the influence of interfering molecular species. Outwardly one gets the impression of vigorous methodological develop- ment and fundamental studies for this technique. Of some 56 publications considered however only three were in refereed analytical journsls whereas the remainder were conference presentations dominated by a limited number of groups.One manufacturer alone gave at least 18 conference presentations some of which appeared to be identical and another leading laboratory gave ten. Some but not all of these abstracts are remarkable in the amount of detail they include and it is possible to obtain a reasonably good understanding of the advances being made. These are considerable and it is only to be hoped that they will become more widely available in full papers. It is often considered that GDMS is the true successor to SSMS and it would appear that GDMS has become such an accepted means of analysis of solid materials that its use no longer necessarily requires particular mention. A new type of glow discharge device has been described and was considered suitable for GDMS (91/719).The device which consisted of a planar cathode in a curved magnetic field relied on gradient drift curvature drift and electrostatic reflection of electrons to achieve high cathodic sputtering rates at low voltages and pressures. The r.f. discharge continues to receive attention. A comparison of the d.c. and r . t discharges (90/C3 129) found that whereas r.f. ion energies varied with both pressure and sampling distance d.c. ion energies spread out only with lowering of pressure. The decline in contaminate gas ion signals with increased changing voltage occurred more rapidly in the r.f. discharge than in the d.c. discharge but both types of discharge showed an increase in 40Ar2+ 40Ar+ and 41Ar+ signal intensities with increasing voltage. Sput- tered ion signals were consistently higher for the r.f.discharge than for the d.c. discharge at comparable voltages but the relative change in the sputtered-ion signal was more pronounced for the d.c. discharge. The r.f. discharge sputtered layers of thin metal films at a rate approximately twice as fast as the d.c. discharge operated at the same voltage. Typical operation of an r.j GDMS instrument has been described by Christie (91/C613) for analysis of both conducting and non-conducting samples and comparison made with a d.c. discharge. Molecular interferences such as metal argides and dimers were common to both sources but more prominent in the r.f. discharge which gave an absolute ion current for the 63Cu+ ion 16 times greater than that given by the d.c. discharge.The r.f. source was a better source of atoms with a sputtering rate 21 times that achieved by the d.c. source but a less efficient ionization medium. The two discharges were considered comparable but the r.f. discharge had the advantage in its ability to analyse both conductors and non-conductors. Sample-aperture separation was found to be a critical parameter for analytical performance of a cryo-cooled discharge source (9 llC2860). An increase in discharge power from 1 to 3 W and 1 kV discharge voltage resulted in an increase of analyte ion currents by an order of magnitude and a decrease in the relative ion currents of non-analyte species by about two orders of magnitude. The best discharge condition for the cell configuration used was 1 kV at 3 mA with an 11 mm sample-aperture separation.Under these conditions the analysis of aluminium on a magnetic- sector double-focusing instrument gave average precision of 3% for relative concentrations in the range 1 x lo-’ to 1 average accuracy better than 10% and detection limits in the low ng g-’ range. Other approaches have been taken to achieve improved performance. The analyte ion current could be enhanced by placing a small positive bias on an isolated element exposed to the plasma (9 1/C2004). Post-quadrupole optics produced an increase in the ion signal and collisionally induced dissociation (CID) removed troublesome molecular inter- ferences. Plasma potential and ion energy both correlated with positive orifice bias (91/C6 10) which indicated that ion energy was defined by the plasma potential itself defined by the voltage placed on the sampling orifice.Harrison and co-workers have investigated use of a pulsed glow discharge. In time-resolved studies (9 1 /C595) enhanced sputter ablation observed in the pulsed mode was compared with that in the d.c. mode. The ion profile for background gas species was significantly different from that for the sputtered species which showed a large ‘afterpeak’. By placing the data gate over this afterpeak it was possible to obtain a spectrum in which analyte ions predominated and the interfering ions were discriminated against. This afterpeak was considered to result from an increase in the population of long-lived metastable Ar atoms which col- lided with the cathode material transferred their energy and produced ions.Introduction of 1% methane as a metastable-quenching agent reduced the afterpeak by 60% confirming that metastable ionization was probably domi- nant in the afterpeak (90/C3 143). These results on metastable quenching provided informa- tion on the ionization mechanisms present in the glow discharge. The three most important processes were consi- dered (90/C3 132) to be electron impact (transfer of energy from electron to sputtered atom) Penning ionization (transfer of energy from metastable argon to the sputtered atom) and charge exchange (reaction between argon ion level and sputtered atom level). The Penning ionization has been studied by Hess and co-workers (91/C604) by intro- duction of methane as a metastable-quenching agent and was found to decrease linearly with increasing current (from 75% at 0.5 mA to 42% at 3.5 mA) but to rise with increasing pressure (from 0% at 40.0 Pa to 60% at 106.67 Pa and above).This clearly illustrates the need to maintain constant source conditions during analysis otherwise ioni- zation mechanisms will change. Jakubowski and Stuewer (9 1/ 1 503) have compared Ar and Ne as working gases and found them to be complementary. Similar analytical performance could be achieved for both gases with optimized operational232R JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY OCTOBER 199 1 VOL. 6 parameters but Ne required a higher gas pressure. A higher pressure resulted in more uniform elemental sensitivities but a lower pressure gave higher sensitivities; hence highest sensitivities were achieved with Ar at low pressures.Use of Ne resulted in mass spectra with few interferences in the important range 50-100 mlz and the low sputter rate resulted in better depth resolution for profiling by GDMS. Matrix effects have received little attention presumably because elemental sensitivities are optimistically consi- dered uniform in different matrices. Two studies have measured relative ion yields for different matrices. Quanti- tative measurement of a number of trace elements in aluminium and copper metals showed that the relative ion yield for some elements (Be Mn and Sn) differed by more than 50% in the two matrices (91lC1655). This emphasized the requirement to match closely the sample and standard matrices for quantitative analysis.The second study (90/C3020) suggested isobaric interferences as the main factor in the observed variation in relative ion yields but no data or results were given. The removal of interferences by CID was highlighted. This use of CID to remove interferences has been studied. Harrison and King (91/C599) have used a triple-stage quadrupole instrument to investigate the identity of parent ions which dissociate to produce the daughter atomic ion. The results of specific collision conditions could also be studied. Duckworth and Marcus (9 1lC603) evaluated the optimum conditions for CID analysis. A cell pressure corresponding to 5 . 3 3 2 ~ Pa in the analyser chamber gave the best compromise between dissociation and loss of analyte due to scattering. At a collision energy of 40 eV interferents were reduced to levels indistinguishable from background noise. It was also noted that detector noise was not reduced under pressurised conditions and the primary source of noise appeared to be from photons emitted from the plasma.Reduction of this photon noise was predicted to result in improved analysis. The same workers have studied target gases for CID (90/C3 127) but no data were presented on the effectiveness of each gas (Ar N2 and Xe) for the removal of isobaric interferences. Source modifications have been made to make instru- ments more versatile. A direct insertion probe is inconveni- ent for the analysis of bulk ceramics so an external sample mount has been constructed to allow the direct analysis of solids of diameters greater than 0.5 in (91/C1893). A minimal pre-sputtering time (2 min) was required and higher plasma powers could be used.A sample holder has been designed for the analysis of thin films for which it is desirable to sputter only the top of the sample (91K606). The sample and substrate approximately 5 mm square were held by a flexible tantalum fitting on to the end of a direct insertion probe and a shield with circular opening restricted the sputtered area. Two lost-cost GD sources have been designed for fitting to the spectrometers used in SCIEX-Perkin Elmer ICP-MS instruments (9 1/C2782). In one system the sample cathode is a central pin and in the other a disc. This has allowed comparison of the GD and ICP as sources for elemental MS but no results were given.Investigations into alternative methods of sample intro- duction have seen some very interesting developments. In particular a liquid chromatograph has been coupled to a GD source via a particle beam interface the excited atoms being sampled by AES and the ions by MS (90lC3024). This raises the possibility of direct determination of metal species separated by LC by both AES and MS. Further details of this procedure are awaited with interest. Harrison and co-workers have used electrothermal vaporization to introduce atoms into a GD source (901C3148). Loading of the sample was a lengthy and time-consuming procedure which involved evaporation of solvent under Ar at atmos- pheric pressure evacuation of the source chamber intro- duction of discharge gas and finally atomization of sample.An improved arrangement of the direct insertion probe was being developed Harrison and co-workers have also investigated laser ablation and fast atom bombardment (FAB) as sources of atoms for GDMS (91K605). Ion signals increased linearly with pulse and beam energies and introduction of a buffer gas at a carefully controlled pressure of 53.33 Pa produced the maximum ion signal. A sharp drop in signal was observed above this pressure due to redeposition and recombination reactions. Non-conduct- ing samples could be analysed directly and preliminary results showed a 10-fold increase in signal for the laser- discharge combination in comparison with a laser alone. It is of note that these new methods of sample introduction are further moves to separate the atomization and ioniza- tion processes and could be considered as special cases of SNMS.Aspects of the application of GDMS to the dnalysis of various materials have been studied. Optimum conditions of discharge power (3 W) and working source pressure (1 x Pa) for the analysis of steel gave a reproducibility of 3-10% and detection limits in the range 2-40 ng g-* (91lC2874). The values of relative sensitivity factor were in the range 0.3-6 which contradicts somewhat the common claim of equal element sensitivity. Relative sensitivity factors obtained for compacted and massive metal samples were almost identical (91lC2868) so could be applied to the analysis of either type. A typical analysis of a pure semiconductor indium illustrated the sensitivity and element coverage of the technique (90K2964).Analysis of tellurium was faster by GDMS than by SSMS but sample preparation was more difficult and detection limits were better by SSMS (9014007). It was difficult to find an internal standard in both techniques but a matrix line could be used in GDMS. Considerable attention has been brought to the analytical performance and possibilities of GDMS using a quadrupole analyser. Typical analyses using an instrument with a cryo- cooled ion source have been presented for the determina- tion of trace metals (91jC2109 91K2856) and low relative atomic mass elements (C N 0 P S and Si) (90lC2963) in metals and alloys. Depth profiling was possible with sputter rates of 0.1-1 pm min-' and detection of 0.5 pm layers (91/C285) whereas thicker layers (5-20 pm) used in galvanized coatings which cannot be analysed by SIMS could also be conveniently analysed (90lC290 1).1.4. Inductively Coupled Plasma Mass Spectrometry 1.4.1. General reviews There have been very few general summary papers on ICP- MS available up to the end of this review year. Gregoire (9 11248 1) in a 5 1 reference review surveyed the use of ICP- MS to determine isotope variations in natural materials such as rocks minerals waters and brines with emphasis on accuracy and precision. The effects of polyatomic ions spectroscopic and non-spectroscopic interferences and mass discrimination on the measurement of isotopes were considered. Reed (9 1/727) discussed recent developments in geochemical microanalysis in a 17 reference review.Bulk analysis techniques based on the ICP (ICP-MS and ICP- AES) were compared with longer established methods (XRF AAS and INAA) including an interesting compari- son of the capital cost of equipment The following Atomic Spectrometry Updates which cover Environmental Analy- sis (901349 l) Clinical and Biological Materials Foods and Beverages (90/3500) Instrumentation (9 l/8) and Industrial Analysis Metals Chemicals and Advanced Materials (9 1 /27 19) included ICP-MS development and applica- tions.JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY OCTOBER 1991 VOL. 6 233R 1.4.2. Fundamental studies 1.4.2.1. Operating parameters and instrument develop- ment. Vickers et al. (9112208) used MS to obtain spatially resolved measurements of ion kinetic energies and densities in an ICP.As a result of changes in the plasma potential ion kinetic energies varied as the ICP was sampled in different locations. These changes were indicated to compare well with those reported by others but caused these workers problems in the interpretation of the measured ion densi- ties. Spatial profiles compared well with those obtained by laser-excited fluorescence in the ICP and axial distribu- tions of BaOH+ BaO+ Ba+ and Ba2+ correlated well with the behaviour of ICP-MS signals that occurred when the inner gas flow rate was varied. Degrees of ionization for As and Sb in an ICP calculated over a range of operating conditions and measured using MS were reported by Houk and co-workers (9 1/2 198). The electron density upstream of the sampling cone varied with plasma conditions in a manner consistent with that re- ported for ICP-AES sources.Variations in ionization temperature of Sb with plasma conditions were consistent with those expected of the ICP itself but this was not found to be the case for As. An He-Ar mixed gas ICP was investigated for use as an ion source for ICP-MS by Sheppard et al. (91/2716). The addition of He to Ar produced a plasma capable of ionizing high ionization potential elements more efficiently than a pure Ar plasma. The best S/N ratios for metallic elements were obtained with a 20% He plasma and for halides with a 30% He plasma. Limits of detection for the non-metals were improved and metal oxide levels were reduced by a factor of two but doubly charged ions were increased by a similar factor.Kawaguchi et al. (9 1 / 7 3) evaluated a water-cooled torch for use in ICP-MS. The behaviour of analytes (Bi Coy In La and Y) doubly charged oxide and polyatomic ions under various operating conditions was examined. Although the water jacket allowed for a low gas consumption compared with a standard torch there was a 50% deterioration in sensitivity and an increase in the level of interfering ions. Ross et al. (9 1/840) characterised a 9 mm low power (850 W) low flow (8.7 1 min-l) torch for use with ICP-MS. This ion source exhibited less dependence of the ion optic bias potential on relative atomic mass than a conventional-sized torch and produced similar sensitivities limits of detection doubly charged and oxide ion levels. However it was found that proper placement of the discharge under the interface was critical to obtaining maximum signal levels for each element.1.4.2.2. Calibration. Semiquantitati3:e (SQ) analysis by ICP-MS is a poorly understood calibration routine that can be very useful for obtaining approximate values for most elements in a sample. The capabilities and applicability of the SQ routine were critically investigated with NIST SRM 1643a (Trace Elements in Water) by Amarasiriwardena et al. (9 1/857). Several instrument parameters and methods of analysis were examined in order to determine their effects on precision and accuracy. The best results were obtained through the division of the response table into three groups of elements and the response of each group was corrected using vanadium caesium or iridium as internal reference elements.Elemental concentrations to within 30% of the certified values could be obtained using this technique. Longerich (90/3289) published a theoretical paper which showed that correction for mass discrimination in isotope dilution ICP-MS is unnecessary if the isotope ratios of the sample the tracer and their mixture are measured under conditions giving the same mass discrimination. The true isotopic compositions of the sample and tracer do not enter the calculations although care in selecting the ratio of tracer to sample is required. 1.4.2.3. Sample introduction. In this review period as in the last research into alternative sample introduction methods contributed the single largest area of published work on fundamental studies.Although there have been many abstracts available for review on laser ablation all except one general review of laser ablation applications in atomic spectrometry (9 I/ 1250) have been from conferences and offer little useful information. Many of these are from instrument companies extolling the virtues of the technique without providing any data. As with laser ablation there have been several reports on the application of electrothermal vaporization sample intro- duction for ICP-MS which reflects the actual amount of research activity in the area but only one written report. Caruso and co-workers (91/856) provided a detailed de- scription of the modifications made to a commercial graphite furnace unit for use with ICP-MS. Ten consecutive firings of a 30 pg sample of Pb resulted in an RSD of 4% for peak height and 1.6% for peak area measurements.The absolute limits of detection were 10 and 14 fg (for a dwell time of 10 ps) for peak area and peak height respectively. The linear dynamic range for Pb extends over three orders of magnitude to 300 pg. Hirata and co-workers have demonstrated the use of a merging chamber with Bi hydride generation (91/2254) Re vapour generation (9 1 / 1430) and spark dispersion of powdered samples (91/835). In the first application Bi hydride was merged with the mist from a nebulizer on the way to the ICP. A flow system allowed a cyclic procedure of introduction of solutions reactions and evacuations to be performed sequentially. The effects of acidity and volumes of sample solution and reductant on the signal were investigated in detail.The absolute limit of detection of Bi was 72 fg corresponding to 1.8 ng ml-l in 40 pl of solution. In the second application pg amounts of Re were deter- mined by introducing sample solution and oxidizing agent into a heating cell placed in the merging chamber. After a drying stage vaporized Re20 was carried to the ICP with blank aerosol sprayed from the nebulizer. The use of the flow system ensured that good reproducibility of signal was obtained. The effects of cell temperature and volume of oxidizing solution on the signal intensity were studied in detail. Using isotope dilution for calibration the Re abundances of four meteoritic samples were determined. The precision of these measurements made using 3 ng of Re was about 1 Yo and the absolute limit of detection was 10 fg almost 25 times lower than that obtained by conven- tional nebulization sample introduction.By using the spark dispersion-merging sample introduction technique the direct multi-element analysis of four geological RM could be carried out on 1 mg of powdered sample loaded on to a discharge cell in the merging chamber. As with the other techniques dispersed sample was led to the ICP with a blank aerosol sprayed from the nebulizer. The relative sensitivity factors (RSFs) of the REE were all found to be similar allowing internal standardization to be used for calibration. However the RSFs for other elements were significantly different from those of the REEs. By using a rock RM for calibration the semiquantitative determina- tion of over 40 elements was performed in a few minutes with an accuracy of k 30%.Absolute limits of detection of 100 pg could be obtained for most elements. An FI system was used with ICP-MS to determine Pt in airborne particulate matter (9013489). The FI system had a cation-exchange resin column on-line for trapping major matrix elements and hafnium which caused suppression of the signal intensity of Pt and a spectral interference by HfO+ respectively. The detection limit was about 0.1 ng ml-l in a simple solution or 5 ng g-l in airborne particulate matter. Observed concentrations of Pt were in the range 14-184 ng g-'. Mochisuki et al. (9112252) used234R JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY OCTOBER 199 1 VOL. 6 FI-ICP-MS to determine Bi Pb and Th in nickel-based alloys.Parameters such as carrier flow rate injection volume and sample concentration had the greatest influence on the sensitivity precision and time of analysis. Samples introduced had up to 0.75% m/v solids content. Results from the analysis of NIST SRMs 897 898 and 899 (Nickel- based Alloys) were in agreement with the certified values. The precision ranged from 2.0 to 8.2% (RSD) and the limits of detection were 1 to 40 ppb. The use of LC for on-line separation of elemental species or removal of matrix components continues to attract interest. Houk and co-workers (Anal. Chem. 1989,61 149) developed an on-line sample treatment method to separate analyte metals from alkali and alkaline earth elements and anions for ICP-MS.The reagent bis(carboxymethy1)dithio- carbamate was used to complex Bi Co Cr Cu Hg Mo Ni Pt and V. The complexes were adsorbed on to a polystyre- ne-divinylbenzene resin column under acidic conditions. The interfering matrix components were not complexed and passed through the column unretained. The metal complexes were then removed from the column with a basic eluent no organic solvent was necessary. The technique was applied successfully to urine and sea-water RMs. Limits of detection were in the range 8-80 pg ml-l. Sheppard et al. (9 1/853) combined ion chromatography (IC) with ICP-MS for the speciation of As compounds commonly found in urine. Ion chromatography was used to eliminate or reduce the mass spectral interference formed from chloride ArCl+ by separating chloride from the arsenic compounds chro- matographically. A 1 + 19 dilution of the urine samples was necessary to avoid column overloading from chloride and subsequent chloride interference.Limits of detection in urine of 3.4 ppb of As"' 4.2 ppb of AsV and 7 ppb of dimethylarsinic acid were obtained. Good agreement with RMs was obtained for total As content which was calcu- lated from the sum of the species. Suyani et al. (9013309) used the ICP-MS as a detector for micellar LC in the speciation of tin. The eluate was passed via an injector (operated as a switching valve) and PTFE tubing to the concentric nebulizer of the ICP-MS which was set up to monitor 120 mlz. Five Sn compounds could be separated with an elution gradient. Limits of detection (as Sn) were between 26 and 126 pg and the response was linear up to 400 ng and the RSDs of Sn at 4 ng were ( 5 % .An on-line flow system for anodic stripping voltammetry was investigated by Caruso and co-workers (9 1/854) for the pre-concentration of Cd and Cu and the separation of Cu from a sodium or uranium matrix. Signal enhancement of up to 50-fold was obtained with pre-concentration of 20 ml sample volumes. Matrix elimination of > 99% was possible for high concentrations of undesired sodium or uranium matrix. Isotope ratio data indicated that the interference of 'j3NaAr+ on 63Cu+ could be minimized. The influence of water loading in the plasma on the signals of ions was studied by Tsukahara and Kubota (91/2214). An ICP-MS instrument with a sample introduc- tion system consisting of a concentric nebulizer a spray chamber and a desolvator (heater and condenser) was used for the work.The amount of water introduced into the plasma was varied by changing the cooling temperature of the condenser. The ratios of Ba2+:Ba+ BaO+:Ba+ and ArO+:Co+ decreased with a decreasing amount of water vapour. Compared with conventional sample introduction oxide and hydroxide ions of REEs were lowered by approximately one order of magnitude with the condenser cooled to 10 "C. Data on ion kinetic energies with and without desolvation were also given. 1.4.2.4. Interference effects. As in the last review year matrix effects have been paid little attention. Caruso and co-workers (91/855) studied the effects of Co Pb and synthetic ocean water matrix constituents on analyte signal.Suppression of the 138Ba+ signal was observed in the presence of each matrix constituent with Pb having the greatest effect. Partial compensation for the matrix induced suppression of analyte signal could be achieved by the use of an internal standard and by initially tuning the ion lenses for maximum analyte signal in the presence of the matrix rather than for a simple acidified standard solution. No significant differences in suppression of analyte signal were found in a comparison of FI and continuous sample introduction. Kim et al. (9 1/220 1) studied the interference effects of 14 matrix elements at concentration levels of 0.01 mol dm-3 on the analyte (Al Co In La T1 and Y) signals. Although a range of parameters were investigated includ- ing mass of the matrix element and analyte ionization potential of the matrix element voltage of the first ion lens and sampling depth no single mechanism could be found to explain the matrix effect and it was concluded that it must be a combination of several mechanisms.Some work has been reported on methods for reducing spectral interferences. Evans and Ebdon (9 1 /852) suggested that the addition of small amounts of an organic solvent (10% propan-2-01 in water) to the sample solution or the introduction of a molecular gas ( ( 5 % O2 or N2) to the carrier gas could reduce the polyatomic ion interference on the As and Se signals. Lam and McLaren (91/850) reported that the addition of 8% N2 to the outer (cooling) gas was a more effective means of reducing oxide interferences in ICP-MS than reduction of the water loading of an all Ar plasma by spray-chamber cooling or partial desolvation of the aerosol.The beneficial effect of the N2 addition was enhanced by combination with reduction of water loading. In addition a dramatic reduction in th ArO+ peak intensity was observed. I .4.3. Applications 1.4.3.1. Geochemistry. A critical evaluation of ICP-MS analysis of geological materials based on case studies was reported by Longerich et al. (9 1/729). The paper discussed a variety of calibration techniques chemical separation and preparation procedures. The evaluated procedures included methods for the determination of 33 trace elements using a modified standard additions technique as well as proce- dures for the determination of Y REE and Th.The application of ICP-MS to isotope ratio determinations was discussed with reference to the determination of 147Sm:144Nd and Pb:Th:U ratios. In a companion paper Jenner et al. (911731) proposed the use of ICP-MS as a tool for petrogenetic studies. Data for 28 elements in seven USGS basalt and andesite RMs were presented. Accuracy and precision for all elements discussed was (7%. The determination of platinum group elements (PGEs) has been discussed by several groups. Denoyer et al. (90/3286) compared ICP-MS with other instrumental methods for the determination of PGEs and discussed developments in sample introduction including laser abla- tion in a review with five references. It was concluded that the accuracy of ICP-MS was comparable to techniques such as ETAAS or ICP-AES and limits of detection for these elements were <5 ng ml-l.Sen Gupta and Gregoire (91/699) determined Ir Pd and Ru in 27 rock ore and related RMs by solution nebulization and ETV-ICP-MS with calibration by isotope dilution (ID) after chemical separation by coprecipitation with tellurium. The accuracy of the technique was assessed by comparison of the results obtained with synthetic solutions and RMs and found to be excellent. A nickel sulphide fire assay and tellurium coprecipitation procedure was developed by Jackson et al. (91/730) for the isolation of the PGEs in geological materials. To correct for matrix and drift effects during the analysis cadmium was used as an internal standard for Pd,JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY OCTOBER 199 1 VOL.6 235R Rh and Ru thallium for Au Ir 0 s and Pt. Over eight analytical runs mean instrumental limits of detection for a 15 g sample were in the range of from 0.7 ng ml-l for Au to 0.07 ng ml-l for Ir. Analysis of the RMs SARM-7 PTC-1 and SU- la indicated that recoveries of PGEs were 86-98% with the exception of Au and 0s. Hall et al. (91/726) reported the development of a method whereby Au Pd and Pt could be determined in 1-2 g of ashed vegetation for practical application of biogeochemistry as a prospecting method. The ashed material was decomposed with HF- aqua regia (HC1-HN03 3+1) and the analytes were separated by coprecipitation with tellurium. Samples were introduced into the ICP-MS instrument by ETV which allowed limits of detection of (0.3 ng ml-l to be obtained at the expense of sample throughput.Measurement of Pd was made at either 106 or 108 mlz because a high concentration of Cu in some samples led to interference on losPd from CuAr+. Accuracy was assessed by comparison of results with other independent methods. Hall et al. (91/707,91/839) reported on the determination of HJ Nb Ta and Zr by ICP-MS and the fusion methods of sample preparation. Compensation for matrix effects arising from the presence of elements such as U was carried out with a ruthenium internal standard for Nb and Zr and rhenium for Hf and Ta. Suppression of the Hf and Zr signals by 1000 pg ml-l of A1 could not be corrected so these workers recommended the use of LiB04 fusion with separation and precipitation of the analyte with cupferron.With this method only 1-3% of A1 is precipitated. Limits of determination of 0.4 pg g-' for Zr and 0.02 ,ug g-l for Hf Nb and Ta were obtained. Results obtained for RMs were in good agreement with certified values and pecision was 24%. Jarvis (91/728) compared LiB04 fusion with an open acid digestion for the decomposition of 23 RMs. The accuracy of the data was found to be dependent on the particular element of interest. Good accuracy (< 5%) was obtained for the determination of REEs Th U and Y in a range of silicate and carbonate matrices prepared by both decomposition methods. Poor accuracy obtained for Hf Nb Ta and Zr was thought to be due to problems in dissolving resistant mineral phases. It was suggested that Pb was lost during the fusion.Precise Rb data are critical for high precision Rb-Sr geochronology. A comparison of ICP-MS and TIMS with ID for the determination of trace amounts of Rb indicated that the former gave better precision by a factor of three (91/2259). Determination of Rb by TIMS precise to 0.6% (one SD) was hindered by uncorrectable fractionation effects whereas fractionation could be monitored during ICP-MS determinations providing results as precise as 0.17% (one SD). 1.4.3.2. Biological and environmental materials. The use of ICP-MS for the analysis of biological and clinical materials is slowly becoming more widespread although severe matrix effects often make the analysis complex. Several groups have analysed human blood serum. Vanhoe et al.(91/1545) carried out a detailed investigation into the problems of determining Co Cu Cs Fe Mo Rb and Zn in serum samples. Indium was used as an internal standard to overcome the effects of 30-40% signal suppression caused by the matrix. A blank solution containing the same concentration of Ca C1 Na and S as human serum was used to correct the results for Co Cu Fe and Zn for polyatomic ion interferences. No spectral interferences were found for the other elements. The results obtained for the seven elements determined in an RM were in good agreement with the certified values. Vandecasteele et al. (9 1/22 19) made a comparison between ICP-MS and NAA for the determination of Sr in a serum RM. Better than 3% precision and a limit of detection of 0.05 ng ml-l was obtained by ICP-MS compared with 10% precision and a limit of detection of 0.02-0.05 ng ml-l by N U .However a radiochemical separation by extraction with 8-hydroxy- quinoline in chloroform was required for N U . Matrix effects from high levels of Ca K and Na were corrected by using indium as an internal standard. Accuracy was further checked using NIST SRM 1577a (Bovine Liver). Lyon and Fell (90/3497) used size-exclusion chromatography to sepa- rate the species which interfere in the direct determination of the 63Cu:65Cu ratio in blood plasma or serum by ICP-MS. An accurate determination of the ratio compared with aqueous solutions could be made directly on the desalted sample. The simultaneous determination of Br and I in diluted ( 1 + 9) blood plasma and urine was reported by Allain et al.(9014 1 59). Limits of detection (calculated as concentration equivalent to twice the standard deviation of the back- ground signal) were 52 and 1.6 ng ml-' for Br and I respectively. Results obtained for samples from healthy individuals were found to be in agreement with literature values. Europium was chosen as an internal standard because its normal concentration in biological samples is very low. This is not a good choice as it is dissimilar in both mass and ionization potential to Br and I. Mulligan et al. (9 1/3) reported on the feasibility of direct analysis of urine diluted 1 + 9 with 2% nitric acid for multi-element determi- nations with limits of detection of less than 1 ng ml-l for urine. The quantitative performance of ICP-MS for Cd Hg and Sb was comparable to that obtained from established techniques.The presence of interferences in the range 10 to 80 mlz prevented the determination of certain elements. Stability of the system over an entire working day was about 5% RSD. Calibration was by standard additions and gallium indium iridium and yttrium were used as internal standards which may explain why no matrix effects were reported. Within the limits described ICP-MS was consi- dered to be a useful tool for monitoring a variety of metals in urine at levels which might be of interest in a programme for monitoring human exposure to chemicals. Tothill et al. (91/1428) determined Pt in animal tissues blood and urine taken from patients and experimental animals up to 3 months after the administration of the platinum-containing anti-cancer drug cisplatin.Following digestion in nitric acid solutions stabilized in hydrochloric acid were analysed by ICP-MS and ETAAS. The limit of detection by ICP-MS was (0.1 ng ml-l and indium was used as an internal standard to correct for any matrix effects. Matrix effects in ETAAS were more severe and the limit of detection was about 10 ng ml-l. A PTFE lined stainless-steel bomb digestion method for human liver and kidney was described by Yoshinaga et al. (91/1485). Samples (0.1-1.4 g) were heated to 120 "C for 3 h. The digest was diluted to 10 ml with water. A 5 ml volume was evaporated to 0.1 ml on a hot-plate at 140 "C and the residue was diluted to 5 ml. The NIST SRM 1577 a (Bovine Liver) results were in agreement with certified values. In a series of papers (91/2314 91/2315 91/2316) the feasibility of using 58Fe as a stable isotope tracer was explored in a study of the absorption of Fe.The under- standing of factors that influence absorption of Fe by infants and children has been impeded by reluctance to use radioisotopes for such studies. In this work it was shown that the 58Fe enriched stable isotope could be used in place of radioisotopes to determine the absorption of Fe from food by infants. Determination of long-lived radionuclides in environ- mental samples by conventional counting techniques can be very time consuming. Work by I(lm et al. showed that 99Tc (91/1084) 237Np (91/1085) and 239Pu and 240Pu (9013434) could be determined in soil samples after a suitable extraction procedure.It was concluded that results obtained236R JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY OCTOBER 1991 VOL. 6 by ICP-MS were in good agreement with those obtained by alpha spectrometry fission-track or NAA but that 1CP-MS offered a limit of detection better by an order of magnitude as well as improved precision and simpler methodology. Igarashi el al. (9 1/225 1 ) described the application of ICP- MS to the determination of both 237Np in soil samples after chemical separation and 232Th and 238U in some biological materials including NIST SRM 1 57 1 (Orchard Leaves) and 1577a (Bovine Liver). Bismuth and thallium were adopted as internal standards to correct for matrix effects in the biological materials. Limits of detection were all t 0 .4 pg rn1-I. The determination of B in both fresh and saline waters was suggested as a possible monitor for industrial pollution (9 1 / 1429). Cation exchange of samples was used to remove high levels of dissolved salts which otherwise would cause a loss of sensitivity and matrix induced mass discrimination. A comparison was made between external calibration standard additions and isotope dilution. For the analysis of prepared water and brine solutions calibration by ID produced the most accurate results. Application of the technique to the analysis of water RMs allowed B to be determined with a precision of 3% at the limit of detection of 0.15 ng ml-l and a precision of 1 % at higher concentra- tions. For the determination of Au in sea-water (91/22) a complex sample handling procedure had to be developed. Samples were preserved with aqua regia on collection spiked with lg5Au recovery tracer mixed with CN- and adjusted to pH 3 with ammonia solution before being passed through an anion-exchange column.The flexible tubing used to transport the solution to the ICP to be pre- conditioned and washed between samples with 8% aqua regia to remove retained Au which could be carried over by the blank solution. With this procedure the limit of detection was 0.08 ng m1-I. Following the assessment of ICP-MS for routine multi- element analysis of soil samples in environmental trace element studies Dolan et al. (91/2503) concluded that in contrast to the single-element technique of ETAAS ICP- MS provided important additional information for a small increase in analysis time.Dry sifted samples were digested with HF-HC104-HN03 in a Teflon bomb followed by the addition of HN03-H3B03 and dilution to volume with water. Isotope dilution was used for the determination of Pb and results for all elements determined in RMs compared well with certified values. Viczian et al. (91/2) concluded that ICP-MS could be applied to establish both the Pb content and isotope ratios in samples such as blood dust paint sediment and soil for identification of potential environmental sources of child- hood Pb poisoning. It was found that sample preparation particularly of soils could influence both the Pb concentra- tion and isotope ratio recovery. Although a variety of sample preparation methods was assessed HN03 treatment in a sealed container with microwave heating was preferred.1.4.3.3. Materials. Rath and Neunteufel (91/1075) in a 23 reference review discussed the use of ICP-MS ICP-AES and AAS for the determination of trace elements in silicon wafers and in the process liquids used in integrated circuit manufacture. Mane is the main gas used in the field of electronics to produce compounds of silicon. Hutton et al. (91/858) showed that the direct analysis of silane by ICP-MS for the determination of elemental impurities at sub-ppb levels was a practical proposition. To minimize the amount of matrix material deposited on the sampling cone orifice an alloy cone was used which operated at a higher temperature than that of the standard nickel version. The optimum carrier gas flow rate with silane was found to be significantly lower than that required to achieve maximum sensitivity in Ar alone.This also reduced sample deposition around the orifice. A further increase in sensitivity was achieved when the Ar carrier gas was supplemented by the addition of H2. The limits of detection were 0.55 and 0.65 ppb for 7 5 A ~ and 12’1 respectively with a precision of 2-5%. Baumann and Pave1 (9 1 / 1 1 1 ) described an analytical procedure for the determination of Th and U in the sub- ng g-l range and of other trace elements in the ng g-’ to pg g.-l range in high-purity quartz samples. The results obtained by ICP-MS were compared with those obtained by NAA FAAS Zeeman-effect ETAAS TXRF d.c. arc OES and XRF. For the ICP-MS measurements the digestion of the samples was carried out with an HF-HN03-H2S04 mixture. Results obtained by the various methods showed good agreement.For Th and U ICP-MS proved to be the most sensitive method with limits of detection of about 50 Zurhaar and Mullings (9 1 / 1 427) evaluated ICP-MS for its applicability to the characterization of window glass fragments with identical refractive indices and of a type frequently presented as physical evidence fo a forensic laboratory. Glass fragments weighing as little as 0.5 mg were digested in PTFE vials using an ultrasonic bath. The concentrations of 48 elements could be accurately determined with a precision of <4% RSD. Analysis by ICP-MS provided a unique elemental fingerprint of a glass sample. Australian window glasses were found to be readily distinguishable from those from the USA The determination of trace elements in NIST SRM 360a Zircaloy-2 has been described by Luo and Chang (9 1 /22 1 1).In order to reduce mass calibration drift the constant temperature control system of the instrument was modified. Accurate determination of Cr Cu and Mn was achieved using external calibration but standard additions had to be used for Hf. It was found necessary to use minor isotopes of Fe and Ti for calibration as there were polyatomic interferences on the major isotopes. Pg g-? 1.5. Laser Microprobe Mass Spectrometry (LMMS) Most of the developments of LMMS have been reported in ‘Microbeam Analysis-1 989 the Proceedings of the 24th Annual Conference of the Microbeam Analysis Society’.A detailed review (9 1/2 1 53) covered historical development and gave typical experimental arrangements. Comparisons were also made with other surface microanalytical tech- niques. Investigations of a fundamental nature are generally aimed at understanding better the processes involved in the interaction between laser beam and. solid surface. In particular emphasis is being placed on understanding errors resulting from non-representative sampling. The shape of craters has been studied using transmission electron microscopy (TEM) and scanning electron micros- copy (SEM) in order to assess the actual degree of re-solidification and accompanying microstructural or stoichiometric changes induced in selected materials (9 112335). A striking difference was found between single- and multiple-shot craters.The apparent depth and width varied with sample type as well as with laser power and focus. In addition material from the bottom of the crater appeared to be re-deposited at the top of the crater tip. Scanning electron micrographs showed a central depression surrounded by a splash zone of re-solidified material. Expansion processes during ionization have been studied and simple models presented to cover the three cases of low medium and high irradiances (9 1/2 162). Poor heat conduc- tors developed a hot spot in the centre of the beam which led to slow decay of excitation after the laser pulse. Good heat conductors on the other hand had a lower maximum temperature better lateral spread of temperature and faster excitation decay.At medium and higher energy irradiancesJOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY OCTOBER 199 1 VOL. 6 237R material was transported across the solid surface. Above a certain energy threshold the so-called plasma ignition threshold the density of the ionized plume increased and plasma absorption became predominant. This resulted in sudden temperature rises and intense ionization. A study of secondary effects in the ionization process also showed that the ion yield was strongly dependent on laser beam intensity with a maximum in the range 1 x to 1 x 10-lo W cm-2 (9112282). The results suggested that the effective region of surface desorption was up to an order of magnitude greater than the geometrical focal region of the laser beam where substrate particles were desorbed.The majority of ions were formed not by desorption but by secondary processes such as Bremsstrahlung of plasma ions and this would impose severe restrictions on the reliability of quantitative spatially resolved surface analysis. Similar conclusions were drawn for the analysis of monolayers (91/2159). Ionization was found to increase with laser power density and wavelength but only slowly with pulse length. Low etch rates were required for monolayer ablation and as a consequence the ion yield was not very representative of the surface composition. This was primarily due to the ion-forming step operating too close to its threshold power and these results led to the conclusion that laser ablation was not particularly satisfac- tory for monolayer work.Although increased power density resulted in a greater degree of ionization the increased ion yields were found to have deleterious effects (91/216 1). A reduction in average ion energy as measured in the TOF tube with increasing ion production was attributed to possible ion-neutral collisions in the denser plasma. Mass resolution peak width and peak shape were all degraded with increased ion production. It was considered that the mechanisms for ion extraction and interaction need to be more fully under- stood. A five-step model was proposed to describe the evolution of peak shape as ion production increases. Pattern-recognition techniques have been applied to LMMS data for standard and environmental particles containing various nickel compounds (9 1 /2 1 67).Ten char- acteristic peaks were chosen to provide fingerprints and the cluster patterns plotted in principal-components space provided the decision rule for the assignment of nickel- containing particles to specific molecular species. The achievements and limitations of LMMS applied to biomedical research have been reviewed (9 1/2 155). Trends shortcomings and new directions were discussed with particular consideration to combinations of different ana- lytical techniques. One interesting combination is that of TLC with LMMS (91/2183). The 60 compounds analysed were all organic but this work suggests possibilities for speciation studies. A study of silica dusts illustrated the power of LMMS when correct instrumental conditions were chosen (91/2182). Aluminium in natural quartz occurs in three forms substituted tetra co-ordinated Si in the crystalline lattice of quartz aluminosilicates and heterogeneous mix- tures of quartz and silicates.At high laser power it was not possible to differentiate between the last two forms because the A1 and Si signals were of equal intensity but at reduced laser power (20-30% transmission) the two forms were distinguishable by the relative A1 and Si intensities. Pos- sible correlation of A1 form to cytotoxicity could then be investigated. At reduced laser power density however some loss of other elemental data occurred. Analysis of individual particles was illustrated by the characterization of individual aerosol particles from the Antarctic (91/2359). The great majority of particles con- sisted of sea salt transformed to various degrees in the atmosphere but small amounts of aluminosilicates (2% of the coarser 2-4 pm range) Fe rich K rich Zn rich and organic particles (all < 1% of the total population) were found.1.6. Resonance Ionization Mass Spectrometry (RIMS) This laser-based technique appears to have reached the stage where the principles are reasonably well understood but the technique is not yet being widely applied on a routine basis. Reported studies fall into two groups one being the building of instruments with RIMS capabilities and the other the continued demonstration of potential applications in particular for the determination of the lanthanides and actinides. Very few true application papers are yet appearing. Resonance ionization is considered in depth in two detailed reviews of laser-ionization spectrome- try (9111134 91/1516).An instrument designed for trace analysis had two atomic sources (9112290). In one the atomic-beam oven the sample was placed in a reservoir and heated by electron bombardment. The alternative was a traditional filament source in which sample electrodeposited on a rhenium filament was atomized thermally. A detection limit of about 1 x lo8 atoms was achieved by step-wise excitation using two dye lasers pumped by a copper vapour laser which was also used for simultaneous non-resonant photo- ionization of the excited atoms. A 40-fold improvement in detection efficiency was gained by performing the ioniza- tion step via excitation of autoionizing states with the help of a third dye laser.This gave detection limits of 2 x lo6 atoms for Pu and 4 x lo6 atoms for Tc evaporated from a filament. Isotopic analyses of Pu were also performed. An instrument with multi-function capability has an arrangement which allows either resonance or non-reso- nance ionization of sputtered neutral atoms in addition to traditional SIMS analysis (91/C626). A 20 pA argon-ion gun (t400 pm spot size) was used to sputter neutral atoms and an oxygen-ion beam was used for SIMS analysis. Each procedure was investigated by the analysis of Ag-Au alloys. The Ag:Au intensity ratios observed by SIMS were up to 300 times greater than the true alloy composition. Those measured by non-resonant ionization were much closer to the true values but appeared to show no correlation to them.Resonant ionization gave a correlation but the non- linearity was still to be investigated. An interesting development has been the use of an ion- trap detector for laser photo-ionization experiments of NO (9 1/C6 16). The ability to store efficiently all ions generated and then mass analyse them offers attractive possibilities. The limit of detection was about 12 ppb which corre- sponded to about 1 x lo6 molecules within the volume of the trap and the time of analysis was 8.3 s. A possible limitation for analysis of more complex samples is the relatively high pressure (0.133 Pa) of the ion trap which leads to collisions. In a TOF-MS modified for RIMS analysis bulk samples were atomized in a furnace (2 cm3) heated by a tungsten filament (91/C629). When loaded with sodium the atom beam lasted 1-2 h.A channel electron multiplier was used for ion detection and examples were given for the analysis of sodium. The resolution of the instrument described was insufficient to resolve all the Cd isotopes and needs to be improved to be of practical use. Potential applications for the RIMS of ion-sputtered neutral atoms (SIRIS) have been discussed (9 112278). Quantitative analysis of semiconductor materials gave low pg g-' detection limits for In in silicon with a 5 pA ion beam current and 30 min integration. The calibration was linear over six orders of magnitude and no matrix effect was observed. High ionization efficiency coupled with high selectivity allowed accurate trace element determinations in238R JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY OCTOBER 199 1 VOL.6 pg sized biomedical samples. Isotope ratio measurements could be made on 265 pg of Cr equivalent to 70 fg of *OCr. Noble gas analyses are feasible with detection limits of a few thousand atoms and detection sensitivity of 1 part in 1 x 1022. Many of the application studies have focused on the lanthanides. Cerium impurities on fused-silica optical substrates could be detected as both free cerium atoms and cerium oxide following laser desorption and RIMS (9 1/23 19). Tunable diode lasers gave partial isotope selec- tivity in the ionization of La atomized on a hot filament but improvements needed to be made to the technique (9 1/C6 15,9 ~2336). Isotope measurements were made with rapid switching of the acceleration voltage and ion count- ing.Poor results were attributed to counting statistics or to time variation in the atom emission from the hot filament. A spectral study of Prn (1 00 ng) and Sm (600 ng) atomized from charcoal on a hot Re filament gave wavelength information for RIMS schemes (9 1/2357). Saturation spec- troscopy with RIMS detection has been used to obtain Doppler-free spectra of lutetium transitions in samples (2 pg) atomized from the side filament of a triple Re filament (91/2354). A procedure has been developed for the chemical separation and detection of trace levels of Tc in environ- mental samples (9 112330). The Tc was electrodeposited on rhenium or tungsten filaments atomized at 2300 K and ionized by a resonant-ionization scheme using three tunable dye lasers pumped by two pulsed copper vapour lasers.Time-of-flight mass analysis and channel plate detection gave a resolution of 2700. An overall efficiency of 2 x for 99gTc corresponds to a detection limit of 1 x 10' atoms. Rhodium has been determined in rocks ores and natural waters by thermal atomization laser excitation of atoms into Rydberg states and ionization by application of a pulsed electric field (9 1/2332). Improved mass resolution in the TOF analyser was achieved by applying pulses with lengths comparable to the time required for ions to pass through the inter-electrode gap. Detection limits were 3 x 1 O-lo O/o in water samples or 3 x 10- l3 O/O following concentration by absorption. A data service has been established at the NIST to publish the necessary information to apply the resonance ionization spectroscopy and RIMS techniques to routine use in analytical chemistry.The first set of data covers As Au B C Cd Fe Ge Pb Si and Zn and presents relevant atomic data appropriate ionization schemes and pertinent opera- tional details for successful RIMS studies (9 1/2 193). 1.7. Secondary Ion Mass Spectrometry (SIMS) This is the most widely used of all the inorganic MS techniques and is particularly popular for materials analy- sis. The name is usually cited in literature titles or abstracts even if no particular emphasis is given to technique development or innovative application and so the number of literature citations is large. Only technique developments and novel applications have been considered for this Update.There have been far fewer reviews this last year but those referred to in previous Updates will still be relevant. Muller (91/2320) surveyed a number of surface analytical techniques including SIMS with particular emphasis on lateral and depth resolutions while more specific reviews were on the potential and limitations of SIMS in electroche- mistry (91/2331) and the analysis of silicon (91/2366). The proceedings of the 7th International Conference on Secon- dary Ion Mass Spectrometry (SIMS VII) have been pub- lished but were not available for inclusion in this Update. It will no doubt give the reader an excellent coverage of all current developments and applications. Fewer new instruments have been reported recently. Cameca have designed a new sector instrument (IMS 1270) intended amongst other applications for isotope analysis (90K2903).The secondary ion optical system included a large magnetic-sector double-focusing spectrometer which allowed simultaneous detection of up to five isotopes. Mass resolution was up to 100000 and the smallest beam diameter was 50 nm (Cs+) or 120 nm (O,+). A new focused ion beam system was unique in that its design although optimized for high current doping lithography and micromilling application made provision for SIMS of sputtered material (9 1/2303). The SIMS data were used for precise end-point detection during micromachining oper- ations. Both secondary ion and secondary electron images could be taken making this a versatile tool. A quadrupole system for microanalysis relied heavily on computer-aided methods for probe positioning r.f.quadru- pole control and image analysis (91/2154 91/2186). A 40 keV heavy-ion probe (typically Ga+ extracted from a liquid metal ion source) could be focused to a spot as small as 20 nm in diameter. The sputter rate was too great for point analysis so the beam was swept across the surface a typical scan of 500 s with a 40 keV 10 pA Ga+ probe (35 nm in diameter) removes one monolayer when scanning an area of 100 x 100 pm. The limits of sensitivity and lateral resolu- tion and problems associated with quantitation were discussed. A limitation of the quadrupole analyser in comparison with TOF or sector instruments is the lack of simultaneous detection which is necessary as information can be removed in one scan of the probe.A scheme was developed for control of the quadrupole to allow simulta- neous acquisition of more than one mass-resolved image during a single scan and thereby approach the performance of the other types of analyser. This procedure permitted parallel mass-resolved imaging procedures and vector-scan microanalysis in which certain features in the field of view could be avoided during the scan. Aspects of quantification of images recorded with a camera-based detection system have been considered (91/2180). The influence of different components of the image formation system (microchannel plate fluorescent screen and camera) on analytically important parameters (linearity of response shading and background) were investigated. The conversion of a grey-scale image into a concentration involved two steps conversion of the grey- scale image into a counts per second image and conversion of the latter into a concentration image.Calculations were carried out on a mainframe computer with user-developed software. A PC-based digital interface for the Cameca IMS-3f commercial instrument has been designed and imple- mented (9112339). The new interface was compatible with the PC architecture and the existing IMS-3f circuitry and implementation allowed other commercially available PC hardware to be used to extend the analytical capabilities. An analytical method for the quantitative depth profiling of multilayer samples has been presented (91/2301). The matrix affinity for oxygen was one factor used in a sputter yield model for correcting matrix-dependent ion-yield variations. The algorithm corrected for both sputtering rate and ion-yield differences in each layer and required determination of quantitative sensitivity factors and rela- tive sputtering rates for all the layers.The depth resolution in the analysis of Cr-Ni thin multilayers improved with either increasing mass or de- creasing velocity of the primary ions (9 112582). This effect was attributed to surface roughness induced by ion bom- bardment. Selective sputtering phenomena in the Ni layer changed ion yield and degraded resolution at the interface. An ion beam with high electron affinity would be an advantage in this instance. The depth resolution in theJOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY OCTOBER 199 1 VOL.6 239R sputtering of single crystals of Ni has also been studied (91/2291). The variance and decay length of the exponen- tial tail of narrow concentration distributions were mea- sured as a function of primary ion energy ion-current density and temperature. The resolution function was an asymmetrical peak whose width was produced by atomic mixing and sputter roughening and whose asymmetry was described by the mixing effect alone. Optimum depth resolution was achieved by minimizing sputter damage through the use of high erosion speed and low sputter roughening. The depth of surface alterations produced by oxygen-ion bombardment of silicon wafers increased with ion-beam energy (9 112306). The altered layer produced within the SIMS pre-equilibrium period was amorphous and was typically 20 nm deep for normal incidence ions at 4.5 keV.Profiling through heavily As doped layers produced segrega- tion of As to the back of the altered layer. Analysis of insulators leads to charge build-up on the surface of the specimen. One method for charge cornpensa- tion is to deposit a layer (30-50 nm) of Au on the specimen surface but this was found to be inadequate for H depth profiling in proton-exchanged Limo3 waveguides (9 1/23 50). Successful analysis was achieved through the use of an electron beam to compensate for charge build-up during depth profiling. This is an established technique for the profiling of semiconductors on insulators or buried insulator samples but this was the first reported application to waveguide material. An exact measurement of the primary beam shape over several decades of intensity was possible by observation of the secondary ion signal during the motion of the primary beam across a strip structure of a suitable combination of materials (9 1/2322).A special mathematical method was used to model the measured primary beam shape of 02+ and Ar+ ions stepped across an aluminium strip on a silicon substrate and this showed that a Gaussian function is not appropriate to define the distribution of primary current. Oxygen-induced profile broadening was evident for Cu and Ga in silicon under 10 keV 02+ bombardment at impact angles of (30" (9 1/2276). The significantly different decays for the Cu and Ga signals were attributed to the difference in impurity transport during beam-induced oxidation.Whereas Cu was found to segregate at the Si02-Si interface Ga diffused uniformly in the surface silica layer. A new model for the calculation of the angle of incidence of low-energy ( t 7 . 5 keV) primary ions improved upon the model used previously by taking into account the effect of the deflection plates for beam positioning (91/2277). The model applied to positive primary ions and a positive secondary-ion detection mode and also showed that the slope of the crater bottom in the x-direction at low primary ion energies was a result of differing sputter yields from one side of the crater bottom to the other. Energy spreads as low as 0.36 keV have been measured for axial beams using new apparatus (91/2266).In addition the total and axial energy distribution of the ions from the ion source and the lenses of the Kornelsen ion gun were measured. A semi-empirical formula was proposed to relate the ion beam diameter to ion energy and ion-source pressure. Self-sputtered Ga has been used as a marker for the presence of low levels of oxygen in the depth profiling of simple Si-Si02 multilayers (9 ~2265). The secondary ion yield of Ga was much enhanced in the presence of oxygen. Other positive ions could also be used as markers. In the case of thicker (7-85 nm) buried Si02 layers various peaks were observed for the Ga+ signal and particular care was required in interpretation. A major problem in the application of SIMS is that of poor quantitative analysis as illustrated by the analysis of Ag-Au alloys in which measured Ag:Au ratios were up to 300 times the true ratio (911C626).Grasserbauer et al. (9 112273) have proposed a strategy for quantitative trace analysis based on the reduction of random and systematic errors (selection of appropriate primary current and secon- dary optical parameters; reduction of artifacts inherent to ion-target interactions; elimination of spectral interfer- ences; and charging and matrix effects) preparation and characterization of suitable reference materials and assess- ment of precision and accuracy. These workers felt strongly that more detailed investigations on sources and magnitude of analytical errors are necessary to be able to make an evaluation of the potential for quantitative analysis.A major limitation to quantitative analysis remains that of pronounced matrix effects. Bombardment of thin (4-33 nm) multilayer samples of Si-Ge with both Ar+ and 02+ beams produced positive secondary ion depth profiles in which pronounced distortions were observed (9 1 /2279). Similar effects were observed in the negative secondary ion depth profiles with Cs+ bombardment. Under Ar+ bom- bardment the distortions were the result of secondary ion yield variations induced by differential incorporation of oxygen into the films during preparation. Under 02+ and Cs+ bombardment the distortions were introduced by differential incorporation of the primary species into the lower sputter yield silicon layers. It was concluded that SIMS depth profiles for thin multilayers were inconsistent with the true elemental composition of the sample.In a study of Sil-,Ge alloys (9112295) the matrix effect on ion yield could be reduced to a negligible level if x=0.005-0.30 energetic (4.0 keV) 02+ beams were used for sputtering positive secondary ions were analysed and sputter rate changes were properly accounted for. The matrix effect became increasingly severe for x>0.30 and decreasing primary oxygen beam energies. Matrix effects could be reduced by using secondary molecular ions for elemental analysis (9 1/2269). The ion yields for AsM- and PM- were greater than those for the corresponding atomic ions M- for dopant elements in GaAs and InP samples. The AsM- ion yield was neither dependent on primary current nor related to the electron affinity of As-.A large enhancement in detection yield was observed for electropositive elements (electron affinity (0.5 eV). Background interfering ions on atomic ions were avoided by using the molecular ions for example the detection limit for Ge was improved by a factor of 100 by using the interference-free lo5GeP- ion instead of the 74Ge- which has an interference from P2C-. The intensity of CsM+ secondary ions formed by Cs+ primary ions also showed a reduced dependence on the matrix. The relationship between molecular ion and atomic ion intensities was linear. Interferences due to molecular ions are ubiquitous in low- resolution SIMS spectra and mathematical corrections usually involve identification of the interfering species and the use of matrix inversion. This however makes the large assumption that all the species contributing to the ion intensities have been accounted for.Alternative techniques which allowed for the possibility of unknown interferences in the data were applied to Pb isotope ratios and gave significantly different ratios to those obtained by matrix inversion (90/2384). This difference was attributed to the inability of the matrix-inversion technique to identify an interference at m/z 205. Background signals owing to organic contaminants particularly prevalent when low ion fluxes are used and sputter cleaning is ineffective were reduced by setting up an Ar d.c. plasma within a TOF instrument (91/2302). Argon gas at 1.333 x Pa was introduced into the system through a leak valve and a - 400 V potential applied to the sample stage to produce a d.c.plasma between the sample stage and extraction plate of the mass spectrometer.240R JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY OCTOBER 199 1 VOL. 6 With a negative bias on the sample stage Ar ions struck the sample mount; with a positive bias they struck. ground surfaces within the instrument and vacuum chamber. The discharge was limited to 5 min duration to avoid sputtering of internal surface materials. A process for depth scale calibration involving the post- processing implantation of a selected atom was described by these workers themselves as being too cumbersome to find broad acceptance for application especially as the improvements achieved are not generally required (91/2280). The method was effective but the implantation conditions such as species flux and energy needed to be carefully optimized for the particular application.Further- more the range of implantation depths was limited to about 100 nm-10 pm and the method has only been applied to simple monolithic targets. The majority of applications of SIMS are to the analysis of pure materials and although considered a routine technique it is apparent from the problems being studied that the instrumental conditions for each particular applica- tion must be chosen carefully. Requirements for depth resolution and ion yield were conflicting in the analysis of Gap-GaAsP layered structures (9 1/2270). The best depth resolution (3.2 nm) was obtained with a primary ion energy of 1.5 keV but below 2 keV the sensitivity factors were observed to be dependent on As concentration and on decay length.Depth profiling has been used to measure A1 concentrations in thin Al,Ga,-,As layers (down to 15 nm) provided suitable calibration graphs were constructed and the measurements were all made under exactly the same conditions of primary ion energy primary ion species and secondary ion optics settings (91/2367). A Cs+ ion beam was more than 10-20 times more efficient than an Ar+ beam for the analysis of Ti & layers and the ion yield was more quantitative (9 1/2267). For quantification under Cs+ bombardment an Au film was deposited on the specimen and ion intensities were normalized to the intensity of 19'Au-. The effect of Cs implantation was also studied. Quantitative analysis could be performed with an accuracy of 5%.Quantitative element and isotopic measurements on natural alkali-borosilicate glasses were made on a double- focusing instrument with a resolution of 10 000 (9 1/2323). Polyatomic ions had a sharper energy distribution than monoatomic ions so energy filtering was used with an offset of 75-125 eV below the acceleration voltage applied to remove molecular interferences. Thin films (about 10 nm) of Au were evaporated on to the surface to conduct away charge build-up during ion bombardment. The potential for high resolution distribution of analysis in metals has been demonstrated by the analysis of tantalum wires (9112262). The high ion-yield made it possible to perform distribution analysis in the ng g-' range with a sample consumption of about 1 pg per data point.Lateral resolution was about 1 pm and sputtering rate was variable over three orders of magnitude to allow optimiza- tion of measurement parameters with respect to sample geometry. The primary ion beam energy window and target temperature needed to be controlled precisely in the analysis of amorphous metallic alloys (9 1/227). The secon- dary ion emission was sensitive to structure and changed greatly and irreversibly during the transition from the amorphous to crystalline state that could occur on heating. Toxicological implications of aluminium have led to a need to analyse tissue samples. In a study of the contribu- tion of environmental A1 to Alzheimer's disease imaging microanalysis revealed the presence of focal accumulations of Al associated predominantly with pyramidal neurones in the cerebral cortex in the majority of renal dialysis patients (90K4035).In contrast such A1 accumulations were absent in control patients. In another study (91/C1715) the release of A1 from pharmaceutical glass containers was lowest for soda lime-silica glass with the surface dealkalized by sulphur treatment. Secondary ion and XPS data emphasized the role of this dealkalization process in affecting the release of Al. The microanalysis of single particles is impressive. The characterization of pre-biological components in interplan- etary dust required class 100 clean-room conditions for sample handling (9 1/2 185). The particles were coated with Au embedded in In and heated to between 150 and 200 "C to form an In-Au alloy.Maps of hydrogen and deuterium ion intensities were acquired alternately with a Cs+ primary ion beam in raster imaging mode and summed to reduce effects of varying ion intensities. Trace element determinations in the low ppm and ppb ranges have been reported for single mineral grains of the iron oxides hematite and magnetite (9112595). Methods for the analysis of minerals and rocks have also been reported. Ion mapping of site-specific concentrations of trace elements in coal is important in considering the environmental consequences of not only the combustion of coal but also the disposal of coal washery refuse and mine spills (9 1/24 14). Complex ion interferences present a problem in the determination of the lanthanides in silicate minerals (90lC4045).An offset applied to the voltage at the maximum of the energy distribution removed most inter- ferences but a peak deconvolution method was still re- quired to separate the Eu and BaO signals. Differences were found in the relative sensitivity factors for different natural mineral and fused mineral standards depending on the crystalline structure and local inhomogeneities. Such prob- lems have been avoided in the determination of the REEs in 13 silicate rock RMs by fusing the sample with Pb2Si04 to produce glass buttons of very uniform matrix compositions (9 11704). In this way SIMS can be used for the bulk analysis of silicate materials. 1.8. Spark Source Mass Spectrometry (SSMS) Each year upon writing this Update one expects to find a decrease in the use of SSMS but the number of reports continues at a steady level indicating that this the oldest of the inorganic MS techniques covered still provides a useful analytical tool for many applications. The number of laboratories publishing articles on SSMS is now very limited. Few fundamental studies or instrumental develop- ments have been reported in the period of this Update.Most papers are application reports which in themselves are not of particular note but they do provide the evidence for continued use of the technique. This is confirmed by the upgrading of two instruments in the former West Germany for continued analysis of semiconductor materials (91/C1738). Ramendik and co-workers (9 112243) continue to develop their quasi-equilibrium model to describe the ion yield in SSMS and laser plasma source MS. Testing of the model with published data on relative sensitivity coefficients confirmed its validity and universal applicability.The model can be used to improve analytical accuracy by application of four elements as internal standards without using standard reference materials. Datta and co-workers have continued their studies of molecular ions produced by the spark source. In a study (91/2284) of the yield distribution of M,+ produced by a multi-isotopic element M they discovered that whereas Ba C C1 K Sb Sr and Te produced abundance distribu- tions that followed the statistical theory Mo Ni and Zn produced distributions which diverged from theory. The effect was particularly marked for Mo and was found to depend on the matrix in the case of silver ions from silver or silver-graphite matrices.Homonuclear molecules obeyed241R JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY OCTOBER 199 1 VOL. 6 the statistical model but heteronuclear molecules were suppressed and an isotopic effect on molecular formation was proposed. This is a remarkable discovery and a warning for isotope studies using molecular ions as it is widely assumed that the distribution of isotopic molecules will follow the statistical model. Further studies are needed to determine how widespread this effect is. The same workers (9 1/2285) found two different distribution patterns for molecular ions of the form of MCn1v2+ or MOn1q2+. One was a monotonic decrease in yield with increasing n and the other a zigzag pattern with maxima at even n.It was concluded that elemental analysis may be in error unless the nature of the molecular mass spectrum for a given matrix is known. These problems have been known for a long time and all workers will be aware of precautions required when using SSMS. Several previous studies have concluded that the source of heteropolyatomic ions in the spark spectrum must lie in the atom-ion recombination reactions in the spark plasma and this has been reconfirmed by a study of barium titanate electrodes mixed with graphite or gold (9112286). The distribution of the ions of general formula BaTiO,+ were found to fit the statistical model. The combination of C,- and Ba2+ ions to form BaC,+ was postulated because the distribution of the last mirrors the distribution of C,- ions.In an interesting development diverse chemometric methods have been applied to SSMS milliprobe data to characterize the homogeneity of a copper standard used for OES (9 1/2263). The proposed strategy included visualization of the data global homogeneity testing for each element testing the anisotropy of elemental distribu- tions search for element correlations and statements on the homogeneity. Some of the approaches used are unconventional. The problems of interferences were highlighted by the analysis of three tellurium standards (90/4007). Multiply charged ions of the tellurium isotopes interfered with many element lines in particular Cu and the molecular ions TeO+ and TeOH+ interfered with Ba and the lanthanides. Multiply charged ions had to be used for the determination of I Mg P and Sb.Detection limits were better however than for analysis by GDMS and concentrations measured for 24 elements were within a factor of 1.5 of those obtained by ETAAS. Isotope dilution remains a popular method for SSMS users. Jochum and co-workers (9 1/295) have achieved precise determinations (4- 12% for concentrations in the range 1-0.0 1 ppm respectively) of Nb in geological samples by improvements in precision accuracy and detection power. The internal standard element Zr was determined accurately by isotope dilution high mass resolution was used for sensitive measurement of low intensity Nb peaks and the element sensitivity of Nb was determined precisely for calibration. Concentrations in the range from 5 ng g-' to 500 pg g-' could be measured by this technique.This laboratory continues to demonstrate the levels of precision that can be achieved by SSMS when particular care and attention are taken. Other typical applications have been reported by the group in Buenos Aires. Antimony determined accurately by AAS was used as internal standard for the determina- tion of Pb (9 1/334). Semiquantitative analysis of cobalt without standards and assuming unity for all relative sensitivity factors gave accuracies (sic) within a factor of 3 (91/2260). The method was used as a survey analysis for the control of the chemical composition of cobalt and the detection limit for Hf (t0.5 ppm) was lower than by other methods. The carbides and oxides of the lanthanides were not observed in the determination of the lanthanides in a glass matrix using graphite as conducting powder (91/2215).This is surprising as previous evidence has always been that oxide formation can be considerable in particular for Nd which can often be determined as the oxide. 1.9. Sputtered Neutral Mass Spectrometry (SNMS) This relatively new technique decouples the atomization (sputtering) and ionization process thereby eliminating or reducing matrix effects which are much more pronounced for ions than neutral atoms. Reviews on the subject testify to the advantages that this decoupling brings and yet most reports are still on technique developments and potential applications and there are few if any reports of routine applications. If the potential of SNMS is borne out then it can be foreseen that it will replace SIMS for many applications but it would appear that commercial instru- mentation needs to be more widely available.The commer- cial instrument available requires separate sources for SIMS and SNMS analyses and some efficiency of SIMS analysis is lost. The potential improvement of SNMS over SIMS was demonstrated by the analysis of Ag-Au alloys in which Ag:Au ratios were much closer to the true value when measured by SNMS than by SIMS (91/C626). The ratios however were only correlated with true concentration when resonant ionization was used so further development was required. A review of several surface analysis tech- niques (9 1/2320) included SNMS and discussed lateral and depth resolution and quantification.The low bombarding energies (< 100 eV) required to sputter neutrals minimized the sputter-induced degradation of depth resolution. In principle the signal intensities contained information to determine the true depth scale. The matrix effect in the determination of As in gallium arsenide was reduced compared with SIMS and SSMS analysis by laser non- resonant or resonant post-ionization (9 1/2272). The Ar sputtering beam was blanked at 100 Hz to synchronize with the repetition rate of the laser. A new type of energy analyser with an electron impact ionizer is compact enough to be mounted in a SIMS instrument to enable both SIMS and SNMS to be per- formed simultaneously or sequentially (9 112326). The analyser has been designed to be sufficiently compact to fit into the existing commercial instrument but this comprom- ised optimum performance.The analysis of an Al-Mg alloy under different partial pressures of oxygen confirmed that SNMS is free from oxygen enhancement and so matrix elements can be quantified with considerable accuracy. The bombarding electrons however caused a space-charge effect which resulted in deformation and shift of the energy distributions of the post-ionized particles. Another similar conversion has been described in which an electron beam post-ionizer was incorporated into the design of an existing system of 'Wittmaack-box' type of SIMS optics (91/2264). The conversion did not greatly reduce the efficiency of secondary ion collection of SIMS analysis. Practical problems encountered in the design of the electron beam post-ionizer included characterization of the energy analyser and conditions required from the optics for optimum secondary ion and residual gas suppression in the SNMS mode. A reproducibility of better than 10% in the relative sensitivity factors and a detection limit of better than 0.0 1 atom-% were obtained for most elemental species in alloys using a 10 pA Ar+ primary beam.Further work is required to characterize the technique particularly with respect to the effect of changes in composition on sputter yield and the subsequent effects on calibration. A multi-function spectrometer system has been designed for SIMS SNMS Auger electron spectroscopy X-ray photoelectron spectroscopy ion scattering spectros- copy work-function measurements mass spectroscopy of242R JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY OCTOBER 199 1 VOL.6 thermally desorbed particles and residual gas analysis within the one unit (90/3438). The sample was rotated between the electron and MS positions within the source chamber to allow faster switching times compared with a linear transfer. This system was limited to solid specimens however. During depth profiling only SNMS gave true depth position whereas the other techniques showed devia- tions. As a single technique SNMS gave the best perform- ance but the combination of SNMS and electron spectros- copy improved the quantification facilities. Typical applica- tions included the depth profiling of multicomponent semiconductor materials. Plasma post-ionization has a higher efficiency than electron beam post-ionization.The Maxwellian electron component of a resonantly excited low-pressure r.f. plasma has been used for post-ionization in a novel high-frequency mode of SNMS (9 1/C2862). A high-frequency potential was applied to the sample against the SNMS plasma so that only an electrically neutral displacement current flowed over the sample which was then negatively self-biasing. A square- wave r.f. potential of appropriate amplitude was used to achieve the same ion optical conditions in the ion- extracting system between SNMS plasma and sample surface as in the direct bombardment mode of SNMS. This resulted in good depth resolution as demonstrated by the profiling of a Ta-Si multilayer structure with a double layer thickness of only 20 nm.The same technique was used for quantitative hydrogen-depth profiling in a-Si:H films (91/2346). The SiH signal was well correlated with a hydrogen content of between 0 and 20 atom-% and so could be used for quantitative determination of hydrogen concentration. A new spectrometer named SNART (sputtered neutral analysis-Riken type) in which an Ar plasma driven by an electron beam was used both for sputtering and for post- ionization has been evaluated (91/2271). The low energy of the sputtering Ar ions resulted in good resolution for the depth profiling of multilayer samples. The sputtering rate (8 nm s-l) was an order of magnitude larger than for SIMS but less than that for GD spectrometry. Quantitative analysis of steel samples demonstrated high post-ionization efficiency depth resolution and sputter rate.No matrix effect on post- ionization efficiency was observed and relative sensitivity coefficients were in the range 0.70-1.15 (compared with 0.26-8.4 for SIMS). The non-resonant photoionization of neutrals sputtered by He gas-moderated laser ablation has been reported (90/2378). A pulsed valve delivered a 100 ps burst of He to the sample surface and the ablation laser was timed to irradiate the sample within 1 ms after the He valve began to open. The vaporization products were directed into the ionization chamber and the ions analysed by TOF spectro- metry. The technique was considered feasible but the complexities of ablation photofragmentation and photo- ionization needed to be better understood.Alternatively non-resonant multiphoton ionization has been used for neutral species produced from alloys by electron-stimulated desorption (9 1 /2274). Ionization efficiency was consider- ably greater than that achieved by electron bombardment and the technique will probably see further development. 1.10. Stable Isotope Ratio Mass Spectrometry (SIRMS) The isotopic analysis of C H N 0 and S by SIRMS is commonplace in agricultural biochemical clinical and geological applications. It is so accepted that rarely does the technique get mentioned in routine application so its use is much greater than is apparent from the literature. Some interesting technique developments and application studies are reported however and this could well increase with the commercial development of GC-combustion-SIRMS which allows analysis of much smaller samples than traditional methods and introduces the prospect of simpli- fied isotopic analysis of individual species or compounds.The gas-chromatographic separation of C 0 2 and N20 improved the isotopic analysis of CO (91/2329). Because the two gases have an identical molecular mass of 44 the presence of N20 introduces an error in measured ratios for C02 and a correction is required if the gases are not separated prior to analysis. The separation employed did not alter values for S13C and P80 in the C 0 2 . At present the sample size is limited to 3 pmol but it is intended to increase this to 15 pmol. The isotopic analysis of soil air can be performed with volumes as low as 100 ml (90/4113).Soil gas was sucked slowly into ampoules taking care that only pore gas and not desorbed gas was sampled. The isotopic analysis of nmol amounts of C02 in the gas was performed by means of dynamic-flow MS in which the sample container holding frozen C 0 2 was coupled directly to the inlet capillary of the MS. Sulphur isotope determination is usually made on SO2 but this method requires a fairly large sample (>4 mg of S) gives fairly poor precision (-t0.2°/oo) and is incapable of measuring some isotope ratios (33S:32S and 36S:32S). An alternative method using SF6 has been developed (91/236 1). The preparation of SF is however complex and strict attention to purity of reagents was required to avoid problems from interfering ions. The SF6 was measured as the SF5+ ion on a mass spectrometer fitted with eight multiple collectors for simultaneous measurement of the ion beams.The method exhibited good precision and accuracy (better than k 0. high sensitivity (sample size 0.04-0.4 mg of S) and the capability of determining all the S ratios. A previously reported method has been improved for the determination of dissolved He in water in the concentration range 4.0 x 1 0-8 to 270 x 1 0-8 cm3 of He (STP) per cm3 of H20 (91/2345). Water samples were collected in 500 cm3 stainless-steel cylinders and the dissolved gases extracted into an evacuated cylinder of equal volume. After addition of 30 psig of N2 containing less than 2 ppb of He to the extracted gases the mixture was analysed for He by MS. Accuracy was 7% for concentrations > 10 x 1 0-8 cm3 of He per cm3 of H20.The analysis of Antarctic snow and ice is highly relevant to the current concern for global warming (90/3557). The deuterium and oxygen isotope ratios of the water molecules are closely tied to temperature and humidity. These ratios provide a method of dating the ice (using the annual summer-winter oscillation) and also reveal a history of longer term changes in climate. The C02 and CH4 content has varied in parallel with temperature over the past 160 000 years. Coupled measurements of 6l80 (by SIRMS) and I4C (by AMS) in planktonic foraminifera have been used to estimate sea-levels at the last deglaciation (9112352). The correlation of age with core depth gave a model to estimate sea-levels back to 12 200 years BP using the 6I8O data.Sea-levels rose by 40 m during the millenium which followed 14 500 years BP. 1.1 1. Thermal Ionization Mass Spectrometry (TIMS) This technique is widely used in isotope geology and nuclear studies and often does not warrant a mention in routine application papers. The basic TIMS technology is well established and few major innovations or advances are reported. Most topics reported cover improved instrumen- tation or new applications mainly using ID techniques with which TIMS is almost synonymous. Instrumentation improvements have been concerned with increasing abundance sensitivity. The VG SectorJOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY OCTOBER 1991 VOL. 6 243R 53-30 is equipped with a 30 cm radius electrostatic energy filter to remove low-energy ions and enhance abundance sensitivity from about 2 to 0.I ppm at rnlz 237 with respect to 238U (90lC3568). Precisions of 0.25% (one standard error) can be obtained for 232Th:230Th ratios of I70 000. A brief note (901C3567) reported a sensitivity of 0.02 ppm for the same measurement on a Finnigan MAT 262 instrument fitted with a decelerating immersion lens and a static quadrupole focusing lens. Typical long-term reproducibility of Sr and Nd isotope ratios have been reported (90lC3563). The maintenance of good vacuum and peak shape and application of a non- power law correction for fractionation resulted in the mean ratios for 87Sr:86Sr in NIST SRM 987 Strontium Assay and Isotopic (nominal value 0.7 10 24) agreeing to within 0.000012. Sample loading for Nd ratio measurement was reduced from 1 pg to 30 ng by using the NdO+ beam instead of the Nd+ beam.Magnesium isotope ratios can be measured by the addition of uranyl nitrate to the sample to act as a binding agent (9 112538). Rhenium filaments were degassed before loading and pre-heated after loading to reduce the back- ground sodium signal. Corrections were needed for electron multiplier discrimination and mass fractionation. Stable isotope tracers of magnesium have been used to measure the bidirectional magnesium ion fluxes in single barnacle giant muscle fibres immersed in Ca and Na free isosmotic media (91/2343). The silica gel technique was used to load 5 pl samples and analysis was by a quadrupole instrument. The striking aspect of this work was the time scale of 15,30 and 60 min over which kinetic constants were determined. This emphasises the versatility of stable isotope tracers in that the tracers themselves impose no limitations on the timescale of experiments High-precision Ca determination is important to be able to search for the small variations in the 4oCa:42Ca which are predicted for whole-rock samples (90/C3 562). Samples were loaded directly on to double Ta-Re filaments in dilute HN03 to avoid the fractionation and contamination effects introduced by phosphoric acid. Large ion beams (approxi- mately 5 x 10-lo A for 40Ca) were measured and the isotopes measured in two groups to cover the wide mass range.Preliminary results indicated an external precision for 4oCa:42Ca ratios of 0.01% for repeated analyses of a CaC03 standard.The use of Ca stable isotope tracers to study kinetics in humans relied on analysis by TIMS using triple filaments and a quadrupole analyser (9 112358). Accuracy and precision for the analysis of standards were 1 and 0.3% respectively and sensitivity in 2 ml plasma samples was 7.5 nmol. Problems associated with high-precision isotopic mea- surement of Li have been discussed (90/4117). At such low masses isotope fractionation is a major source of error. Double filaments (tantalum for the sample and rhenium for ionization) were loaded with 3 p1 of 1 mg g-l boric acid solution and LiOH sample solution containing from 10 ng to 1 pg of Li. The conversion of LiOH into Li2B407 had to be complete otherwise excess of LiOH would ionize and add to the fractionation.The presence of anions also increased fractionation. Lithium was determined as the Li+ ion which gave a sensitivity three orders of magnitude greater than that for the Li2B02+. Contamination by Na required the use of higher temperatures which sacrificed precision and at levels above 100 pg of Na a broad peak for m/z values of from 5 to 23 masked the Li isotopes. Replicate analysis of a standard gave an RSD of O.O23%. In the analysis of LiCl loaded on triple filaments (Ni for the sample and W for ionization) a large dependency of measured ratio on filament temperature was noted (90/4119). Between 1 150 and 1850 K a relative variation of more than 1% was found in the isotope ratio. It is to be noted that the sample loading (50 pg of LiCl) was very high and could in itself have had an effect on the analysis.An ionization filament with higher electron emission energy Re for example could have been used to improve analysis. The difficulties experienced in the determination of Rb are centred on fractionation which cannot be corrected directly as there are only two isotopes. Separation of Rb in a pure form is difficult yet the presence of other cations affects the fractionation as does the amount of sample loading. Replication data have shown that uncorrected fractionation which was not reproducible from sample to sample accounted for 80-90% of the scatter of results (9 112259). It was suggested that ICP-MS was more suitable for the determination of Rb. Total evaporation of the REEs eliminated isotopic fractionation which was the main cause of systematic error (9 1lC1804).The method already developed for the deter- mination of U and Pu was used to measure Gd Lu Nd and Sm in nuclear fuels. An ID procedure exploited the different thermal ionization properties of the REEs to reduce isobaric interferences to negligible levels (90/C3564). A complete analysis was obtained in less than 3 h on a five-collector instrument with analytical precision of 0.1 Yo and reproducibilities of 0.5%. An external precision of 0.01% was achieved for the determination of Ce by careful choice of fractionation laws and isobaric interference corrections (90x3 5 73). The measurement of Th and U is fundamental in both isotope geology and nuclear studies. A two-tracer (233U and 235U) ID method gave two independent values for the determination of U (using 5-10 pg loadings on double filaments) (9 1/242).Parameters such as chemical separa- tion procedures mass fractionation and the heterogeneous distribution of U in minerals were evaluated. A method has been described for the measurement of 230Th:232Th ratios in mid-ocean ridge basalts which is particularly difficult because of the low ratio (5 x and low Th concentration (0.3 ppm) (91/23 18). Typical samples of 200 ng loaded on to triple rhenium filaments gave a beam of >0.4 x 1 0-l1 A for at least 2 h. The modified spectrometer had two magnetic sectors in an S configuration and an ion counting detection system to give abundance sensitivities of Direct analysis of solids by conventional TIMS is uncommon because high-precision work requires careful control of the loading procedure and of the instrumental parameters. In a method for the high-precision isotopic measurement of B direct fusion of boron carbide (1 pg) samples with Cs2C03 was carried out on tantalum filaments previously loaded with a graphite slurry (9119).The addition of graphite enhanced ionization and improved measurement precision by at least a factor of two. The B was measured as Cs2B02+ at rnlz 308 and 309 rather than as the previously reported Cs2B407+ which gave a weak and unstable beam. Precisions of better than 0.03% were achieved. Measured loB:l1B ratios were about 0.05°/o lower than values obtained by conventional fusion procedures. The highly precise analysis achievable by TIMS makes it the method of choice to establish relative atomic masses.Antimony has been measured on a canoe-shaped rhenium filament loaded with one drop of a mixture of silica gel sample as SbC13 and 100% H3P04 (9112363). At a filament temperature of 1450 "C a beam of 5 x 1 0-l2 A was obtained for the SbO+ ion and simultaneous measurement at mlz 137 and 139 gave precise isotope ratios. The accurate relative atomic mass of 121.7575-tO.0009 (30) adds one more significant figure to a previously reported value of 121.757k0.003 and is further evidence that the accepted value of 121.759 should be revised. Isotope dilution by TIMS is considered the definitive method for elemental analysis and Heumann and co- to 7 x 2 x 10-7.244R JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY OCTOBER 1991 VOL.6 workers (9014003) continue to develop procedures for the determination of Se by negative ionization. Acid digests of sediments and related samples were spiked with enriched 82Se and the Se separated by formation of SeH in a hydride generator. Sample loading on double filaments was silica gel (10 pl) containing the Se sample and Ba (30 pug) as Ba(OH)* to decrease the electron work function of the filament material and to increase production of negative ions. This improved the sensitivity by a factor of over 40 compared with previous methods and 0.5 pg of Se gave an ion current of 1 x lo-" A. The detection limit was 6 ng g-l with a precision in the range O.8-4.l0/o for concentrations in the range 0.2-3.5 pg gel. The same group have determined Cd Cry Ni Pb and Zn in pure copper as part of a certification campaign (91/2261).After dissolution of the sample in aqua regia (HC1-HN03 3+ l) Pb was separated from the matrix by anodic electrodeposition the other elements by anion-exchange chromatography. Positive thermal ions were produced on single filaments using the silica gel technique with phosphoric acid for Cd Pb and Zn and with boric acid for Cr and Ni. Detection limits were 0.03,2,4 1 and 13 ng g-l for Cd Cr Ni Pb and Zn respectively. The Carious tube technique has been used for the dissolution and spiking with 34S of a number of metallic standard reference materials and the S concentration was then determined by TIMS (9 1/826). 1.12. Other Methods A new instrument with potential for surface analysis with atomic resolution has been built by replacing the aperture and single-ion detector of a conventional short-flight-path atom probe with a position-sensitive detector with a larger acceptance angle (9112365).Individual atoms of the high- purity material were removed by field evaporation through the combination of a d.c. voltage and high voltage (1-2 kV) pulses with ns rise-time. Both the chemical identity and surface position were obtained to give not only atomic layer depth resolution but also sub-nm lateral resolution. Ramendik and co-workers continue to develop their quasi-equilibrium model to describe the ion yield in the laser plasma source MS. The model for the description of ion formation has been tested and found to be valid and universal (9 112243). The kinetics of ion creation were used to estimate the conditions under which ionization processes within the laser plasma occur (91/2247).The effect of these processes on the relative sensitivity coefficients was demon- strated and a theoretically based approach used to select laser spot diameters for quantitative analysis. Relative sensitivity coefficients calculated from the kinetic model agreed well with those measured for elements with ioniza- tion energies of between 4.5 and 9.5 eV in geological materials using a laser with 8 x lo8 W cm- energy flux and 150 pm spot diameter (91/2248). The innovative application of resonant laser ablation led to an enhancement of several orders of magnitude in the matrix signals for AlGaAs semiconductor and calcium metal (91/2289). Both A1 and Ga were resonantly excited by the same dye laser.The ionization conditions for Ca were similar to those observed in RIMS indicating that atoms were ionized in the vapour state and that surface bonds were not an important factor. The proposed model for ionization involved the formation of a plume of neutral atoms and some ions followed by ionization of the neutral atoms by photons from the same laser pulse. A further development would be the replacement of the ablation laser in LMMS with a tunable laser. The potential of Fourier transform MS (FTMS) remains elusive and only one group that at the University of Metz France appears to be active in developing the technique for inorganic analysis. A laser microprobe FTMS was built with a central laser focalization lens surrounded by inverted Casegrain optics (90/2375).This configuraton had the advantage of sample visualization a laser beam perpendi- cular to the sample surface (to give homogeneous energy distribution and the possibility for precise defocused beam experiments) and different light paths for laser and visualization optics which permitted laser post-ionization of neutrals. Ions were performed in the source cell resulting in no loss upon transfer. The technique was applied to the analysis of semiconductors and was shown to have a high resolving power with all the interferences resolved (9 1/238 1). An entire mass spectrum was acquired with one laser shot and accurate mass measurements with a spatial resolution down to 5 pm were made. A review of ion cyclotron resonance spectrometry has been published with a considerable delay since the period covered 1983-1986 but the discussion of principles instrumentation and ionization methods is still relevant (9 1 /228 7).Quadrupole storage or ion trap MS is another technique with possible potential for inorganic analysis which has yet to receive attention. A recent book brings together all the pertinent information on the quadrupole ion trap and gives a detailed account of the underlying theory of its operation (9 1/2344). A review of recent developments placed particu- lar emphasis on experiments in which ions are generated in an external source and injected into the trap for separation and detection (9 1/2257). All applications considered were to organic analysis but combination with SIMS using a Cs+ ion beam was included.The major limitation is the effect of space charge effect on the number of ions that can be held together but the relatively inexpensive and versatile instru- ment is capable of fast and accurate analysis. Mass spectrometric techniques traditionally associated with organic analysis also find limited application to inorganic analysis. Insufficient ion yield limits FAB to only a few elements but the technique has been used for the isotope analysis of Mg in biological samples (91/C668). A fast atom beam has also been investigated as a source of sputtered neutrals for GDMS (91K605). Ion yield for Cu increased linearly with beam energy (4_8 keV) and was in the range 50-500 counts s-' compared with 2000-12000 counts s-l obtained with laser ablation.Aggarwal and co-workers (9 112256) continue to develop GC-MS methods for isotopic analysis in clinical and nutritional studies. Various chelating agents were investigated for Cr Cu Ni and Zn and whereas the procedure appeared promising for the determination of Cr and Ni at ppb levels it could only be used with caution for Zn and not at all for Cu. The method was proposed as a more readily available alternative to TIMS or ICP-MS but is severely limited by the lack of suitable chelating agents poor accuracy and problems associated with metal exchange within the GC system. 2. X-RAY FLUORESCENCE SPECTROMETRY 2.1. Reviews tion for a 'New horizons' review (911203). In addition to Gilfrich who has over many years at the Naval Research presenting the importance of synchrotron sources the Laboratory Washington DC USA been in a unique properties and potential of laser-generated plasmas and X- position to investigate new XRF instrumentation and ray lasers as X-ray sources were indicated.These devices techniques concentrated on developments in instrumenta- are in the early stages of development and not yet producingJOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY OCTOBER 1991 VOL. 6 245R X-ray emissions of sufficiently high energy to be of practical use. The use of layered synthetic microstructure (LSM) optics and TXRF were briefly covered and the combination of appearance potential (AP) XRF and EXAFS was presented as uniquely revealing in the simultaneous study of structure and composition of materials.The conclusion on data reduction was that there is unlikely to be any significant development in the 1 990s-the increasing appli- cation of pattern recognition expert systems and chemome- trics may yet prove this proposition to be premature. An assessment of the relative merits of PZXE in compari- son with various nuclear and atomic spectroscopic tech- niques (NAA XRF and TXRF) has been made by Maen- haut (9111565). Many criteria were considered but it was clear that the increasing use of TXRF and synchrotron radiation (SR) XRF and continuing improvements in conventional XRF were eroding the popularity of PIXE in its traditional areas of strength such as biological and environmental analysis. However it was concluded that PIXE was still the only technique to offer non-destructive multi-element analysis of mg sized samples with sub-pg g-l detection limits.The claim that PIXE yielded a lost cost per analysis providing that the equipment had already been built and in almost continuous use for other purposes was rather disingenuous. The review by Reed (91/727) on recent developments in geochemical microanalysis indicated that conventional electron microprobe analysis was in a mature state and that PIXE and SRXRF microprobes were capable of providing concentration detection limits better by one and two orders of magnitude respectively. The last two techniques analyse a much larger sample volume than when electron excitation is employed and this can be regarded as either a benefit or disadvantage depending upon the nature of the analytical problem.Ure (9 1/25 12) included studies using XRF spec- trometry in a review on the determination and speciation of trace elements in soils. The biennial X-ray spectrometry fundamental review in Analytical Chemistry (9 112455) was packed with 337 references spanning the period 1988-1989 in its very extensive but necessarily rather bibliographic coverage of the entire XRF Jield. The authors excluded all publications in Russian and Chinese which accounted for 14 and 12% respectively of XRF papers in 1988. This balance was partly redressed by Ma and Yuan’s wide-ranging Chinese review (90/347 1) covering 373 Chinese references and the reviews in German by Tolg (91/1148) and Janssen (9 1/2440). Scimeca (91/1133) has reviewed the technique of variable exit-angle XRF spectrometry and described its use for the determination of the very light elements ( Z t lo) especially in thin films.Jenkins (9 1/ 1004) has presented an extensive review containing much of the material to be found in the 1989 book entitled ‘X-ray Fluorescence Spectrometry’. The previous MS-XRF Atomic Spectrometry Update was pub- lished in October 1990 (9111436). 2.2. Instrumentation The Royal Society of Chemistry Instrument Criteria Sub- Committee of the Analytical Methods Committee has prepared Part VI of its series ‘Evaluation of Analytical Instrumentation’ (91/2231). The subject of Part VI was Wavelength Dispersive Spectrometers and a methodology was proposed for comparing the features of sequential WDXRF spectrometers. The paper was presented in heavily commented tabular form with weighting factors which were proposed to assist a prospective purchaser to evaluate and compare fully the range of available systems.The proposed scoring scheme was intended to make the instrument selection process more objective and quantitative but assumed that the decision to use a sequential spectrometer had already been arrived at. Hill et al. (91/8) included a section on X-ray techniques in their Atomic Spectrometry Update on Instrumentation. In future the Instrumentation review will be restructured and all XRF work will be covered in the MS-XRF Update. The technique of X-ray microfluorescence (XRMF) has again been widely reported during the review period. An X- ray microprobe with a beam spot of 10 x 10 pm which was in operation on an SR beam-line at the National Synchro- tron Light Source (NSLS Brookhaven NY USA) has been used for elemental microanalysis of thin-section biological samples (911352).Detection limits were around 10 or 3 fg when using filtered and collimated direct SR excitation or the newer Kirkpatrick-Baez focusing optics respectively. This very high sensitivity and small beam size made the SRXRF microprobe particularly valuable for the non- destructive in vitro mapping of trace element distributions in thin tissue samples. Jaklevic et al. (9 11204) have used the same SRXRF microprobe to analyse a 50 pm mark on an antique piece of paper. The detection limit for Fe was 3 fg and a scan of the mark at 5 pm intervals allowed an element map to be built-up which indicated that the Fe concentra- tion reached 130 pg cm-2 at the centre of the mark.Jones et al. (9 1/115) have described the characteristics of the X-ray microscope (XRM) which is installed on the X26 beam-line at NSLS. These workers described a wide variety of biological and geochemical applications and one of the more exciting developments was the possibility of a chemical state X-ray microscope (CSXRM). Baryshev et al. (90/3538) reported the scanning XRF analysis of polished thin-section samples of rocks and minerals using the VEPP- 3 (Novosibirsk USSR) SR source. A pyrolytic graphite monochromator was used to tune the beam over the energy range 8-35 keV and a spatial resolution of 60 pm was obtained. If the direct continuum beam was used then the spatial resolution was improved to 30 pm which is similar to the best figures quoted for laboratory systems using collimated X-ray tubes.Using a measurement time of 1-3 s per point two-dimensional maps of Fe and Sr distribution were built up and a detection limit of 10 ppm was obtained. Although the high brilliance and natural collimation of SR sources are an obvious benefit for XRMF work much valuable work can still be achieved with a laboratory X-ray source XRMF instrument. Carpenter and co-workers (91/201) have designed and built an XRMF instrument based upon a high-brilliance micro focus X-ray tube (gener- ally operated at 30 kV and 0.7 mA) with a Pyrex capillary tube inserted through the normal tube window position. The X-rays from the radiographic microfocus X-ray tube experienced total reflection from the inside walls of the capillary and when one end of the capillary was placed close to the anode focal spot and the other end close to the specimen a small intense spatially uniform beam spot was produced.These workers described the effect on intensity and spectral distribution of using capillaries of 10 and 100 pm i.d. and indicated that owing to the relative intensity gain observed with increasing capillary length capillaries as small as 5 pm i.d. may yield sufficient beam intensity for practical high-resolution XRMF work. The utility of the system was demonstrated (9111589) by the production of high-resolution element maps of an Fe-Ni-W alloy 2% zirconia in yttria and Fe and Mn in a banded agate sample. An XRF mapping instrument has also been described by Kobayashi (9 1/ 1054).The instrument used a conventional X-ray tube and a pinhole collimator to provide a spatial resolution typically of 200 pm. Maps for two elements over a 20 x 20 mm area of rat skin and leaves were produced. Further work to use a fine-focus tube and an X-ray waveguide in an instrument similar to that of Carpenter246R JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY OCTOBER 199 1 VOL. 6 (91/201) was predicted to improve spatial resolution to 30-40 pm. In their study of a new method for the preparation of X- ray optic eZements based upon carbon-tungsten and car- bon-nickel layered synthetic microstructures (LSM) Arba- oui et al. (90/4174) have developed a new sputtering system. These workers were able to control the deposited layer thickness with an accuracy of better than 0.01 nm and the quality of the devices was demonstrated by the high measured reflectivity for carbon K X-rays.Fujisaki and Nakagiri (9014 1 7 5) have designed a gradient refractive index phase zone plate which was made from two materials the concentration of one increased and that of the other decreased gradually with increasing radius in each pair of zones. A ‘lens’ made to such a design from chromium and titanium had an efficiency of 34% which was a great improvement over the typical value of 10% measured for Fresnel zone plates. The new device was expected to find use in soft X-ray microscopy of biological specimens. Wittry and co-workers (91/114) have discussed the avail- able crystals for making focusing elements for use in laboratory-source XRMF instruments.The aim of their work was to provide focusing elements with high collection efficiency which can be used in conjunction with a titanium target for the trace microanalysis of elements with Z< 2 1 in biological materials. The high-resolution X-ray image intensiJer described by Hailey et al. (9 1 /2592) had a particularly large area (1 2 cm2) but the reported spatial resolution of 300 ,um at 22 keV is unlikely to generate much interest for XRMF despite the high X-ray stopping power and fast current-detection electronics. Lundon (91/2547) and Hetala and Kalnicky (91/145) have reviewed the application of on-line XRF and XRD analysis. Areas such as mineral processing in which on-line XRF anal ysers have operated successfully for many years were identified and the increasing trend to place XRF analysers in the actual production environment was noted.The design of a new combined XRD and XRF on-line ore- stream analyser was presented in detail and its application to analysis of apatite was shown to be successful (91/145 and Ahonen et al. Adv. X-ray Anal. 1989 32 59). The design of an on-line XRF system for zinc-lead ore slurry analysis was briefly described by Holynska et al. (91/144). The flow cell was mounted in a by-line and was fitted with a Melinex window a lo9Cd radioisotope source and an Ar- filled gas proportional counter. Harding (Harding A. R. Adv. X-ray Anal. 1989,32,3 1) has described the instrumen- tal characteristics of an on-line analyser incorporating a Peltier-cooled Si(Li) energy dispersive X-ray detector.The detector and X-ray tube were mounted in a conventional. geometry and results for catalyst solutions plating solutions and loose powders were presented. An improved version of this system in which a higher-power water-cooled X-ray tube was used and up to eight flow cells could be fitted has been presented more recently (91/C209 1). Peresie and Chenoweth (90/C3026) developed an on-line XRF analyser which used a lo9Cd radioisotope source and an NaI(T1) scintillation detector to determine the Sr:Ba ratio in television picture tubes. The rapid nondestructive analysis was used to identify tubes from different manufacturers so that these tubes could be successfully segregated during the manufacture of televisions.The problems encountered during the calibration of an on-line XRF instrument were claimed to be greatly simpli- fied by the use of fundamental parameter (FP) absorption- enhancement correction procedures (9 1/143). The Fp method allowed the number of actual calibration standards to be minimized without any notable loss of accuracy for the determination of Br and Pb in unleaded gasoline and additives in lubricating oil. A fully automated sample preparation and XRF analysis scheme has been developed for the routine quality control of raw mix in a cement works (9 I/ 1223). The whole process from receipt of sample through sample grinding and pelletizing to loading and analysis in the sequential WDXRF spectrometer was automated and results were transmitted to the computer controlling the plant.A jeld-portable XRF analyser incorporating a radioiso- tope source and a sealed gas-proportional detector for use in alloy sorting was described by Piorek (911152). The sorting algorithm was based upon a pattern recognition technique using data for up to ten elements and was claimed to be highly effective for all common types of alloy and also to be suitable for sorting of some carbon steels. The ability to sort out stainless steels differing only in their S concentration was demonstrated to be a major benefit of the pattern recognition method over the conventional calibration and FP methods. For quantitative analysis a modification of the Lucas-Tooth and Price intensity correction model was used. A portable WDXRF analyser has been used for the determination of trace amounts of Au and W at levels down to 6.pg g-l in an ore body (91/1404).Bisaro (90/3533) has designed and built a double crystal monochromator capable of positional accuracy of 0.01 s of arc. The goniometer was driven by a combination of stepper motors and piezoelectric crystals and its angular position was determined by means of a Michelson inter- ferometer. Arai and Omote (9013540) have made a detailed study of the intensity and distribution of background in WDXW spectra. They calculated the relative proportions of Thomp- son to Compton scattered radiations and reported the characteristics of a shelf and a long tail on the long wavelength side of major peaks in the spectrum. The analytical advantages of using LSM ‘crystals’ for the quantitative determination of light elements by WDXRI spectrometry have been shown (9111 592).Peak intensities when using LSMs for the determination of F Mg and Na were improved by a factor of 5 15 and 25 respectively. A simple method was proposed for improving the accuracy by correcting absorption-enhancement effects and analysing just one similar analytical standard. Appearance potential XRF analysis (APXRF) has been described in a paper by Kirkland and co-workers (90/3 5 3 9). The very high degree of tunability which can be achieved through the use of a double-crystal monochromator and an SR beam are necessary pre-requisites for such measure- ments and are not commonly available. The ability of APXRF to provide local atomic structure and elemental composition data simultaneously is particularly useful where X-ray techniques contribute to the complete charac- terization of materials and led these workers to propose an instrument configuration which could perhaps be con- structed using a fairly conventional laboratory X-ray source and optics.* 2.2.1. Excitation Much of the published work during the review period relates to the use of SR sources although there does appear to be research continuing into the determination of light elements using somewhat more conventional instrumenta- tion. The improvements in performance when using a scan- dium anode X-ray tube in the determination of light elements in steel by WDXRF spectrometry have been reported (9013364). For the determination of Al C P S and Si the sensitivies obtained with the scandium anode X-ray tube were higher by a factor of 2 than those obtained with a rhodium X-ray tube and the spectrum background was comparable.Accuracy for the determination of C in high-JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY OCTOBER 199 1 VOL. 6 247R speed tool steel was shown to be 0.023% over the range Brummer and co-workers (9112650) have used a high- power X-ray tube and a Barkla scatterer to investigate improvements in detection limits available with polarized excitation EDXRF spectrometry. When compared with direct excitation the peak to background ratios in the energy range 5-40 keV were improved but this improve- ment was in part due to the limitations of the instrumenta- tion in terms of low processed count rate and limited selection of X-ray tube high voltage and primary beam filter.The techniques of low-energy electron-induced X-ray spectrometry (LEEIXS) is of particular interest for the analysis of surfaces and determinations of the very light elements ( Z t 10). Gaillard and Romand (9112437) have described an LEEIXS instrument which was equipped with a cold-cathode source and wavelength dispersive spectro- meter. The excitation system produced a quasi-monochro- matic electron beam whose energy was selectable over the range from 0.5 to 5 keV and which could probe a sample area of about 1 cm2 to a depth of 5-200 pm depending upon the incident beam energy and the type of sample. The spectrometer was used for the near-surface analysis of insulators and some metallic oxides of Cr Fe Mn Ni and Zn.A similar spectrometer was claimed to yield sensitivity 20 times higher than conventional WDXRF spectrometers and this resulted in a detection limit of 30 pg g-l for B in glass (911C1806). Musket and Holmes (91/2394) have designed and built an ultra-high vacuum low-energy X-ray spectrometer the key components of which were an aluminium-anode end- window X-ray tube fitted with a 2 pm aluminium exit window and operated at up to 10 kV and 60 mA and a Si(Li) detector fitted -with an ultra-thin vacuum-tight window of 200 pm Parylene-N coated with 200 pm of aluminium. Careful consideration was given to obtaining a highly uniform beam spot of high intensity. The output spectrum from the X-ray tube was measured by direct observation of the source using a Si(Li) detector and aluminium foils between the tube and detector.The spectrometer was used specifically for the determination of 0 on and in materials such as beryllium graphite alumina sapphire silicon copper and other metals (9112395). A detection limit of 98 ppm (atomic) 0 in beryllium was obtained after in situ ion sputter cleaning of surface 0. As the atomic number of the material increased the emergence depth of oxygen K X-rays decreased rapidly and the detection limit of 0 in silicon and tantalum was 0.39 and 4.3% (atomic) respectively. Jaklevic and co-workers (9 1 /204) have presented a comparison of SR and conventional X-ray excitation sources and demonstrated the benefits which accrue from the tunability of SR. The high excitation flux available from the SR source offsets the intensity losses resulting from the use of a monochromator and other optic elements which are positioned in the beam line.These workers used a high- resolution double-crystal monochromator to provide highly monochromatic incident radiation which could be finely tuned over the energy range 5-20 keV. When measuring the K emission lines detection limits for Mn Rb Se and Zn in a 20 mg cm-2 cellulose matrix were expected to be optimized by tuning the energy of the incident radiation to a value immediately above the K absorption edge of the particular element of interest. In fact this was found not to be the case owing to increased spectrum background resulting from multiple incoherent scatter of the incident radiation.The lowest detection limit was obtained when the excitation energy was tuned to about 1.2 times the K absorption edge energy. In a further demonstration of this extreme tunability the severe spec- 0.76-1 -23% C. tral overlap of Pb La (10.549 keV) on the As Ka line (10.532 keV) in a sample containing 460 ng cm-2 of Pb and 2 ng cm-2 of As was eliminated by setting the energy of the incident beam to 12.9 keV. At this energy the As K absorption edge (1 1.863 keV) was exceeded but there was no excitation of the Pb L lines as the LIII edge energy (13.044 keV) was not exceeded. Despite these obvious advantages of tunability the intrinsically multi-element nature of SRXRF analysis does tend to be lost in this mode of operat ion. Gilfrich (9 11142) has described the characteristics and experimental set-up of the seven synchrotron facilities at which SRXRF work is in progress.The use of both continuum and monochromatic SR was described and a graphic comparison of photon flux output between the Stanford SR source and conventional X-ray tubes was presented. The polarization characteristics and tunability of the SR were discussed and a trend towards use of wavelength dispersive detection systems in place of the more usual energy dispersive detector was identified. In the determination of Ba and Ti for which there is overlap of the L and K lines respectively the energy of the incident beam was tuned to excite only the Ti K lines and so permit quantitative measurement of Ti in a 20 1 excess of Ba. This experiment serves to indicate the lengths to which some workers will go to compensate the inadequacy of their peak deconvolution software which should readily be able to cope with such a case of spectrum overlap.The experimental set-up at the LURE facility (Orsay France) has been described (9013430). A crystal monochro- mator and geometry which maximized the benefits of beam polarization were used in the excitation system. The results demonstrated the improved detection limits and accuracy obtained for a wide variety of sample types including archaeological relics geological materials alloys and food additives . 2.2.2. Detectors One area showing particular growth during the review period has been that of charge coupled devices (CCDs) which are now becoming available with response character- istics suitable for use in the X-ray region of the electromag- netic spectrum.These devices consist an n x m array of charge storage sites or pixels which are potential wells formed under suitably biased electrodes on a silicon wafer structure. Once the charge has been deposited by exposure to X-rays the array is read out electronically by moving the array of charges (pixels) line by line to a read-out line which is then read out and measured pixel by pixel. Thus the position of the original pixel and the (X-ray) energy deposited at it are determined to yield spatial and energy information simultaneously. Lumb (Lumb D. H. Nucl. Instmm. Meth. Php. Res. 1990 A290 559) has described the instrumental character- istics of CCD detectors and proposed a convenient method for calibrating them.A CCD array with 1152x770 ele- ments each 22 pm square was used to produce a detector with an active area of 450 mm2. The detector was cooled to 180 K and read out at approximately 35 000 pixels s-l and yielded energy resolutions of 80 eV (full width at half maximum FWHM) at 0 eV and 160 eV at 6.4 keV (Fe Ka) which are comparable to many Si(Li) detectors. These workers noted the need for very low noise amplifiers and described work which was aimed at improving detector efficiency at energies >2 keV by increasing the depletion depth of the detector. Currently available devices have depletion depths of 4-6 pm which results in very low sensitivity for X-ray energies > 5 keV and the proposed use of higher resistivity silicon was expected to lead to a concomitant increase in efficiency for higher energy X-rays.248R JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY OCTOBER 199 1 VOL.6 A group from Neuchatel published a two-part series on the general description (9 1 / 1 566) and detailed technical aspects of the electronic read-out system (91/1567) of commercially available CCDs for use as low-energy X-ray detectors. The principal requirements of this group were for a detector with good spatial and energy resolution and ability to reject high energy prays and neutrons. The experimental configuration and the characteristics of the 576x385 pixel CCD array and its read-out control and measurement electronics were described in detail. For these experiments the spatial resolution given by the 22 x 22 pm size of each pixel was more important than the energy resolution which was determined to be 236 eV FWHM at 5.9 keV.The detection system was used in the successful study of 2.43 keV X-rays from pionic H atoms in a large- scale physics experiment. Improvements in energy effici- ency and resolution will benefit this type of study and could lead to more general use of CCD detectors in X-ray spectrometry although the correction of low-energy tailing on peaks due to ‘spreading’ of charge over more than one pixel will need careful consideration in order to permit a S/B adequate for XRF analysis of real multi-element samples. Clarke (Nucl. Instrum. Meth. Phys. Res. 1990 A291 1 17) has summarised the characteristics and demonstrated the immense value of combined spatial and energy informa- tion available from CCD detectors in synchrotron radiation X-ray spectrometry. In one study a CCD was used as a position-sensitive detector for measuring glancing angle X- ray reflectivity of platinum-carbon multilayer mirrors. In a second study the circularly polarized portion of the SR beam and a CCD detector were used to measure the magnetic effects on the X-ray near-edge absorption spec- trum of thin-layer samples of cobalt and a cobalt-gold superlat tice.Although the review by Campbell (91/1564) is entitled ‘X-ray Spectrometers for PIXE’ there is much of interest in it for XRF spectrometry with energy dispersive detectors. In particular this worker’s research group at Guelph has concentrated on developing models to explain Si(Lz) detec- tor characteristics such as line shapes and the silicon dead layer which are important both in the understanding of the detection mechanism itself and for EDXRF spectrum processing. The existing models of the silicon dead layer were critically examined and a model which better explains the detector response at low energy presented. The detector line shape models which the Guelph group have proposed included terms which separate out the various features arising from incomplete charge collection observable on the low-energy side of the full energy peaks.These models may be used to improve peak-fitting methods particularly in trace element analysis where modelling or correction of the spectrum background is critically important. The use for alloy analysis of a mercury(zz) iodide detector exhibiting an energy resolution of 588 eV (FWHM at 5.9 keV) has been described (90/3437).The resolution com- pares unfavourably with the 130- 140 eV obtainable from a contemporary Si(Li) detector and the Cr Ka and yB doublet was not resolved at all. Peak deconvolution software using measured library peaks was advocated as there is no peak model available to account for the observed severe tailing on the low-energy side of the peaks. Accuracy of analysis for the major elements in a number of NIST high-alloy steels was in the range 4.7-10.5% relative which is inadequate for quantitative analysis but forms a basis for alloy sorting. The analysis time of 5 min necessary to obtain the above accuracy is probably a consequence of the small detector area and will need to be much shorter if as recommended the system is to be used in a hand-held alloy sorter. Despite increasing commercial interest in high-purity germanium (HPGe) detectors only one paper (9 111 39 1) has described the use of such a detector. The detector was used for the analysis of stainless steels but there were no specific details of detector performance or of how the problems of correcting for Ge K escape peaks were overcome.In such an analysis the Ge K escape peaks from Mo and Nb K lines fall in the energy range 5.7-8.7 keV and will cause a significant interference with the characteristic K lines from Co Cr Fe Mn and Ni all of which are to be found in stainless steels. 2.2.3. Total reflection XRF (TXW) Klockenkaemper (9 1 /2400) has reviewed the essential characteristics of TXRF instrumentation and analysis.This very readable account had examples of ultra-trace water analysis and of recent application to the analysis of biomedical thin sections. The main instrumental features and equations describing TXRF have been described in detail by Schwenke et al. (9 1/1588). The salient equations were presented along with informative plots of how the angle of incidence influences X-ray beam . reflectivity energy distribution and surface penetration. The optimum instrumental configurations for thin film and surface analy- sis or depth profiling were described in detail and expected detection limits were quoted. Yun and Bloch (91/2172) have also presented the requirements for surface analysis and depth profiling using TXRF although their instrumen- tation was based upon a monochromatic SR source and was rather confusingly described as X-ray near-total external fluorescence (NTEF).Their application however was novel in that they studied metal ion concentrations in a liquid monolayer in situ in a Langmuir trough. The present techniques used for sample preparation in TXRF analysis have been reviewed by Prange and Schwenke (9 11148). The extensive experience of these workers of a very wide variety of sample preparation techniques particularly those for matrix removal were succinctly described with many practical illustrations of the achievable results. The interesting use of TXRF for element oxidation state speciation has been presented by Mukhtar and Haswell (9K756).Initial work has involved the measurement of yB/Ka line ratios and there appears to be some possibility for the speciation of As and Cr in waters. The use of TXRF for forensic analysis is particularly attractive as it offers high sensitivity non-destructive analysis and analysis of low mass samples. Toshio and Taniguchi (91/2648) have reviewed their use of TXRF in forensic investigations of a wide variety of materials including cloth plastics lipstick and blood. Ninomiya and co-workers (9 1/26 14) in the determina- tion of As have demonstrated the importance of controlled drying of the liquid droplet on to the quartz reflector. In order to obtain a uniform analysis spot the reflector was treated with a fluorocarbon polymer in much the same way that other practitioners use a siliconizing fluid to form a hydrophobic surface on the reflector.The extreme impor- tance of restricting the sample to a small central spot in order that the whole spot was in the area seen by the Si(Li) detector was noted. This would have been somewhat less of a problem had an 80 mmz detector been used which is the norm in the majority of instruments. The use of TXRF for semiconductor wafer surface analysis continues to be an area of growth (9 1/1194 9112589). Total reflection TXRF instruments are now advocated for quality control in wafer manufacturing plants where they are used for process development and for determination of surface contamination during processes such as solvent cleaning annealing and implantation. Of particular note is the non-destructive nature of the tech- nique and the capability of some instruments to performJOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY OCTOBER 1991 VOL.6 249R depth profiling in addition to surface contamination analy- sis. Detection limits are typically 10" atoms cm-2 although improvements in certain systems through the use of rotating-anode X-ray tubes and monochromators have been suggested (9 1/ 147). Detection limits (1 000 s) better than 5 x 1Olo atoms cm-2 for Co Cu Ni and Zn on silicon wafers have been claimed by Sako and co-workers (91/1591 91/2398) for an instrument in which the volume around the silicon wafer sample was evacuated and additional incident beam filtration was used in order to reduce scattered spectrum background. Further reduction in scattered back- ground intensity was achieved when the X-ray tube voltage was kept at 30 kV rather than the conventional 50 to 60 kV.2.3. Calibration and Data Processing As computing power increases and becomes more widely used in the XRF laboratory there is an increasing chance to investigate what chemometrics can offer the XRF spectro- metry community. Collaboration between the Graz and Freiberg groups has resulted in the development of a prototype expert system based upon a PROLOG inference engine for qualitative EDXRF analysis (91K1823). In an extension to this work a more general neural network approach has been adopted and this was predicted to yield a general method for automated qualitative analysis in WDXRF EDXRF and ICP-OES spectrometry (911C1826). The use of a partial least squares (PLS) technique to replace conventional empirical calibration in XRF analysis has been proposed (9 1 / 1 5 36).The procedure was applicable where matrix matching was possible and had therefore similar requirements to the classical empirical procedures in terms of the number of standards. Accuracy was improved when the intensity data were preprocessed to eliminate measurements with poor precision. The Rasberry-Heinrich data set was used to test the method and results were encouraging. The PLS method was claimed to be better than either conventional empirical or FPT methods for these Cr-Fe-Ni alloy standards and further work will investigate application to other systems in which severe absorption-enhancement effects are encountered. Kaneko et al.(9 1/ 12 17) have extended the idea of using log(energy) versus log(sensitivity) plots for calibration in the determination of trace elements in refractory borides carbides and nitrides. The extended equation log (I/C)=A(log +Blog(E) +D has a square term includedand was claimed to improve accuracy over the previous form of the equation when applied to the cited analyses. The basic log@) versus log(sensitivity) calibration procedure was also used by Yap and Vijayakumar (9 1 / 1 5 36) in order to minimize the number of standards required for the determination of 14 trace elements in Japanese pottery. The quantitative results were used in a principal components analysis (PCA) to determine the geographical origin of the samples. Of the 14 elements determined only six (Mo Nb Rb Sr Zn and Zr) were actually required to group the samples.The PCAmethod was superior to the ternary diagrams which Yap had applied previously and could see more widespread use although the quoted analysis times of 10000-50000 s per sample are unlikely to be used elsewhere. De Boer (91/1557) gave the detailed derivation of the equations used in the calculation of XRF. intensities from both single- and multi-layer thin film samples. A modifica- tion was proposed to the equations in which the very important inter-layer enhancement could be derived analyt- ically and which therefore was more amenable to com- puter evaluation. In fact the analytical form of the calcula- tion was found to be 5 5 times faster than the previous method of numerical integration for the calculation of iatra-layer enhancement and six times faster for inter-layer enhancement.A comparison of the method of LaChance and Traill (LT) using theoretical alphas and the empirical alphas method of DeJongh has been made (9013270). The LT method with theoretical alphas was recommended for its ability to yield an initial analysis based upon a single standard in which accuracy of a few percent. was achieved and accuracy equivalent to the DeJongh method when multiple in-type standards were used. Rousseau (9 111 586) has described the use of empirical and theoretical alpha techniques and shown how the latter led to the use of a fundamental algorithm which combined the use of theoretical alphas with a simple and universal calibration procedure.The operational details of a program which implements this fundamental algorithm have been summarized (9 1/ 1 587). For the analysis of geological powder pellets Ji and co- workers (9 111 568) found that the theoretical alphas method using only ten standards gave more accurate results than either an empirical concentration or intensity correction model using 50 standards. Feret and Fortier (91/C1721) reported impressive accuracies (much better than k 0.01%) in the determination of elements at concentration < 1% in a wide range of aluminium alloys when applying theoretical alphas calculated by the XRF-11 program. Eggert (91/1553) has presented an overview of the equations currently used for the correction of Si K escupe peaks in EDXRF spectra. A new two-parameter equation was propposed and a Turbo Pascal routine to implement a channel by channel escape peak stripping algorithm was presented.The calculated escape ratios agreed with those of Statham (J. Phys E 1976 9 1023) and the channel by channel algorithm was also applied to the stripping of yB peaks in order to correct for spectrum overlap. However the often severe differential absorption between the Ka and I@ peaks in EDXRF spectra are likely to militate against the use of this algorithm in EDXRF spectrum processing although its use in electron-excited systems could be considered. Wood and Quarles (90/1856) have studied the effect on the K XRF intensity of S by calcium X-rays in a calcium sulphate matrix. It should come as no great surprise to find that the study revealed the necessity for a correction of the enhancement ofS by calcium in thick (10 mm) samples but not in thin film (2-11 mg cm-2) samples of the same composition.It is perhaps timely to remember that not every X-ray spectroscopist has immediate access to contemporary desk- top computer power. Torok (91/1558) has described four very simple but useful compact programs which run on a pocket programmable calculator. The programs performed conversion of wavelength or Bragg angle to energy and computation of Bragg angle and resolving power of a specific instrument configuration. Finally the use by Galloo and Guillermo (90/3949) of 695 samples for the calibration of an XRF method for the determination of Pb in atmospheric dust is remarkable and must surely be the antidote to 'standardless' XRF analysis.2.3.1. Fundamental parameter (FP) calculations A new FP program which was claimed to offer standardless analysis has been developed for WDXRF spectrometers (9 1/1201). The program used an extensive sensitivity library which appeared to be transferable from instrument to instrument in order to allow standardless analysis. The library allowed for the use of five different analysing crystals and two detectors. Itoh et al. (9 112583) have reported the use of the XRF- I I FP program for the traditionally difficult analysis of nickel- based heat-resisting alloys. When one calibration standard was used the accuracy was almost as good as that which was achieved when theoretical alphas and 15 Calibration skan-250R JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY OCTOBER 199 1 VOL. 6 dards were used.The reduction in the number of standards required was clearly a major benefit of the FP approach. A semi-fundamental parameter calculation method has been reported in Japanese by Ochi and co-workers (9 1/268 1). Unfortunately publication in a more accessible form must be awaited before any assessment of the novelty accuracy and utility can be made. 2.3.2. Matrix correction using scattered radiation A simple Compton scatter correction procedure for the determination of the heaviest element in a light matrix has been used for the determination of Cu in organic copper compounds (91/74). An extension of this procedure was developed (9 1 / 1 3 73) and successfully used for the determi- nation of Nb Zn and Zr in laboratory-reagent grade rutile and Zr in glass. The extended procedure corrected for absorption edges between the analyte line and the Compton peak and was based upon mass absorption corrections from Compton scatter measurements and at an effective wave- length on either side of the intervening edge.Tanaka et al. (91/1609) used the linear relationship between mean atomic number of an aerosol sample and the Compton/Rayleigh scatter peak intensity ratio to correct for absorption effects. The mass absorption correction was calculated from the mean atomic number in a manner similar to the backscatter fundamental parameter method. The proposed procedure was claimed to provide accurate results for the determination of light elements (Al Ca and Si) in aerosol samples.2.4. Applications 2.4.1. Specimen preparation Agrawal and Kapoor (9013269 91/1552) investigated heter- ogeneity effects which occurred during the preparation of in-house calibration standards for the determination of trace levels ((1500 ppm) of Ca Gd Th and Y in an alumina matrix. Two preparation methods were used one in which analyte compounds were dry mixed with micro- crystalline cellulose and the other in which solutions of the analytes were absorbed on cellulose and subsequently ignited at 800 "C. The method of dry mixing was subject to the greater error and if it was used it was recommended that the effective mass absorption coefficient of all added materials should be similar. This is obviously rather difficult to achieve in practice so the sorption method is to be preferred.Tao et al. (9 1/201) have described how XRF analysis can contribute analytical information during the production of geological and biological SRMs and certified RMs (CRMs). The extreme importance of homogeneity testing for such standards was rightly stressed and a timely reminder was given to users of SRMs and CRMs that the minimum amount of standard recommended on the certificate must be used if the certified values are to be applied. For example the minimum amount for NIST SRM 1573 Tomato Leaves is500 mg and this should be rigidly adhered to. Radioisotope-excitation EDXRF has been used for the checking of homogeneity of Fe Sr and Zr in a lake sediment RM (91/146). Sample sizes of 20-50 mg were taken and analysed directly and the automated non-destructive na- ture of the analysis was a particular benefit. Alvarez (9 1/2396) has investigated the preparation of glass fusion beads in the light of theory and practice from the glass-making industry.The theory concerning removal of stress in the glass by annealing was described in detail and polarimeter measurements were made in order to determine the amount and types of stress introduced into fusion beads. The annealing process was studied by differ- ential thermal analysis and an inexpensive simple method for producing glass discs using a muffle furnace and graphite crucibles and moulds was described. A low-dilution fusion method for the determination of major minor and trace elements in a wide range of rock and mineral samples using a single bead has been developed by Eastell and Willis (91/202).The choice of flux oxidant and dilution ratio was investigated and a 2:l dilution compris- ing 2 g of dried sample 4 g of Spectroflux lOOB (80% lithium metaborate-20% lithium tetraborate) and 0.6 g of lithium nitrate was recommended. The lithium metaborate helped to make the melt more fluid and the flux was found to be suitable for a wide variety of silicates and also for more-difficult samples such as kimberlites and lamproites. Higher sensitivity compared with 'normal' dilution and the elimination of particle size and mineralogical effects com- pared with pressed pellets were the benefits claimed for the method and particular advantage was claimed for the determination of Ba Cr and REEs. An even-lower dilution (0.5 1) fusion method has been developed for the analysis of Portland cement (9 1/149). An 80% lithium metabo- rate-20% lithium tetraborate flux with no additional oxidant was used in a 0.5 1 flux to sample ratio with HBr as non-wetting agent. The beads required rapid cooling to minimize crystallization then grinding and polishing to obtain a smooth flat analytical surface.The low dilution minimized any deterioration of detection limits in the EDXRF measurement but required the use of an FP matrix correction procedure. A procedure for the fusion of sulphide-containing miner- als concentrates mattes and drosses has been described by Norrish and Thompson (911205). The method involved a short pre-oxidation stage using sodium nitrate prior to fusion. Losses of S and migration of metals in to the platinum ware were sufficiently small to allow the method to be used routinely.The importance of maintaining oxidizing conditions and stirring the melt during fusion has prompted Couture (9 I/ 1 5 1) to design a new fusion apparatus. The equipment incorporates a gas burner and also a laboratory mixer to agitate the melt. In order to exclude the reducing atmos- phere of the burner plume from the sample air was pumped through a nickel tube inserted into the crucible lid. The setting of the angle of the crucible and the manner of its rotation were important and were described in sufficient detail for the apparatus to be produced in other labora- tories. The procedure was claimed to be almost 100% successful in forming perfect beads and no problems were noted from Fe reduction during fusion.Beads were cast into a special flat mould which was easy to maintain flat even with repeated usage. Izquierdo and Guevara (90/C4105) found difficulty in obtaining platinum ware equipment and reagents for fusion and recommended its replacement by the briquetting of samples with high-purity wax binder. Although data were broadly comparable between fusion and briquetting methods of preparation the pressed pellet method loses the critical advantage of removing sample history enjoyed by fusion. The use of pressed pellets should only be regarded as a replacement for fusion when insur- mountable difficulties prevail in carrying out fusions. Ochi et al. (9 1/ 12 1 8) have designed a sample cell which allowed the analysis of liquids in a vacuum environment.The cell had a 15 mm diameter 50 pm beryllium window which it was claimed prevented expansion of the liquid and formation of bubbles thereby allowing its use for the analysis of liquids in top-view WDXRF spectrometers which is usually rather difficult. The cell was also fitted with an inner film of Mylar and the lightest element determined was P although reproducibility of sample loading was quoted only for solutions containing Co Fe Mn Pb Sm and Zn the measurement of which does not require aJOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY OCTOBER 1991 VOL. 6 251R vacuum path. In the determination of S in oil the problem of analysing the liquid was circumvented by mixing the oil with paraffin wax at elevated temperature and cooling to form a solid disc (9 1/426).A detection limit of 20 ppm of S was obtained from a WDXRF spectrometer but care must be exercised not to irradiate the pellet for too long or there is a danger of the wax softening and even melting in the heat generated by the X-ray beam. The combination of hydride generation and WDXRF analysis has been reported for the determination of As Sb and Se in biological samples (91/228). A 1 g sample was digested overnight with nitric acid (1 0 ml) then heated with perchloric acid (2 ml) before cooling and dilution with water and coprecipitation of the analyte elements with lanthanum hydroxide. This precipitate was then separated and placed in a reaction vessel where the hydrides were generated and collected on a cellulose filter impregnated with silver nitrate solution in a central spot defined by a wax ring.The working range of the method was 0.13-20 pg g-l for As and Se and 0.03-20 pg g-l for Sb. Methods of preconcentration and separation continue to be studied although rediscovery of the utility of precipita- tion methods for the preconcentration of trace metals from water (91/1580) could be avoided by consulting the exist- ing extensive literature on the subject. Sodium diethyldi- thiocarbamate was used for the precipitation of trace amounts of Se from 11 samples of waste water (91/1055). Recovery was >97% and the detection limit was 1.9 ppb. Kobayashi and Nakamura (9014 165) used precipitation by 1 -( 2-benzothiazolylazo)-6-bromo-2-naphthol (BBTAN) for the preconcentration of Co Cu Fe Ni and Zn from waters and digested biological samples.Despite the use of thin-film specimens of filtered precipitate the high intensity of the Br K lines from the BBTAN caused substantial enhancement of the analyte K lines andan alternative reagent had to be found. Hauer and Wegscheider (91/C1762) have described the use for preconcentration of trace elements in water of dithizone and dibenzyldithiocarbamate (DBDTC) on re- versed-phase columns. Of particular note was the develop- ment of an automated system to control column loading elution and conditioning. A digestion and preconcentration method for the determination of trace concentrations of Nb and Ta in rocks and minerals by WDXRF has been described (91/1159 91/2545). The powdered sample was digested with 1 + 1 HF+HC104 and the resulting solution was treated with N-phenylbenzohydroxamic acid solution in toluene.The extract was evaporated on to 0.5 g of cellulose powder which was dried and pressed on a boric acid backing to form a 25 mm diameter pellet. The detection limits of 10 and 20 pg g-l for Nb and Ta respectively are rather high considering the lengthy prepa- ration and may be a consequence of high scatter from the boric acid backing or of the instrumentation employed. A method for the preconcentration of the PGE and Au using a fibrous polymer sorbent containing Polyorgs X 1 -H which contains benzimidiazole groups has been developed (9 1/2242). After application of the solution containing the PGE the sorbent was dried ground and pressed to form a pellet 4 mm thick and 40 mm in diameter which was then analysed using a sequential WDXRF spectrometer.The analytical range was 7-300 pg g-l. It is disappointing to note that Kolmogorov et al. (90/2185) reported their XRF methods for the determination of trace elements in geologi- cal samples only to be ‘competing with AAS and ICP-OES’ even though they used preconcentration methods and SRXRF. Favier et al. (91/C652) have demonstrated the use of ion- exchange column chromatography for the separation of As species prior to EDXRF analysis. The chromatographic method was used to separate arsenite arsenate monome- thylarsonic acid and dimethylarsinic acid. Portions ( 10 pl) of the column eluate were deposited and evaporated on a 4 pm thickness Mylar support film then analysed using an in- house EDXRF instrument.The spectrometer consisted of a high-power molybdenum anode X-ray tube whose output was strongly filtered by a thick molybdenum foil and a Si(Li) detector placed close to the specimen. Detection of pg amounts of many elements was claimed and detection limits of 10 ppb were obtained for the routine determina- tion of the As species. For the quantitative analysis of milligram amounts of non-metallic material Lemm et al. (9 1/ 1205) used a pressed pellet method. The sample typically 100 mg was blended 3 + 1 with wax placed in a die and then gently compressed. In order to make a robust pellet a backing made from a mixture of 5 g of iron(rI1) oxide and 1 g of wax was then placed in the die and pressed at a force of 6 tonne. The iron(rrr) oxide was added to increase the mean atomic number of the sample and thereby reduce spectrum background intensity.The pressed pellets were used for the determination of 51 elements in 60 min by a sequential WDXRF instrument. 2.4.2. Thin films Willis (90/3274) has presented a very detailed study of the inter- and intra-layer enhancement effects which can occur in multi-layer thin-film samples. The intensities for various layer arrangements compositions and thicknesses of samples containing Fe Ni and Cr Cu Fey Ni layers in which severe enhancement effects occurred were calculated. One particularly interesting example quoted was the case of a nickel layer on a tin substrate where the Ni intensity was 125% of that expected for a bulk nickel specimen.For the analysis of multi-layers such as those containing Cr Fe and Ni in which enhancement is strong an error of 1OOOh in calculated mass thickness can occur unless the enhance- ment is properly corrected. A review on the subject and thin-film analysis in silicon wafer technology included coverage of TXRF in its seven pages and 48 references (9012337). The use of a sapphire substrate instead of silicon for the measurement of P in borophosphosilicate and phosphosilicate glass passivation layers was claimed to improve precision from 0.32 to 0.05% m/m (91/1194). The improvement resulted from decreased interference by Si K lines from the substrate use of peak intensity ratios instead of absolute intensities and elimina- tion of wafer reloading effects. 2.4.3.Geological materials Hall et al. (9013586) have used data from 1600 samples in a comparison of instrumental NAA (INAA) sequential WDXRF ICP-OES and AAS methods for the determina- tion of 12 elements in rocks. The precisions obtained for the determination of Be Fe and Na by XRF were better than those obtained by INAA. In general the XRF and ICP-OES were found to offer significantly better accuracy than INAA for the elements concerned. Norman et al. (90/2190) compared ICP-OES XRF INAA and isotope dilution MS methods for the analysis of silicate rocks. The results showed excellent agreement between the techniques and the use of ICP-OES for routine silicate analysis was rec- ommended. It should be noted however that the ICP-OES work for the comparison was carried out in these workers’ laboratory and that data for the other instrumental tech- niques were obtained from other institutes.The techniques of EDXRF ICP-OES and ETAAS have been compared for their suitability to the determination of Sn in a variety of geological materials (9 1 /294). Only EDXRF spectrometry could analyse the powdered samples without the need for sample preparation and only EDXRF and ICP-OES yielded252R JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY OCTOBER 199 1 VOL. 6 acceptable accuracy over the range of materials encoun- tered. The detection limits using ICP-OES and EDXRF were 5-10 pg g-l and 2 pg-l of Sn respectively and the EDXRF method appears to offer the best choice for this analysis. Eller et al. (91/1502) used the techniques of GFAAS and TXRF in complementary combination for the ultra-trace determination of Au Os Pd Pt and Rh in rocks.These workers presented an extensive review of the sample preparation methods available for separation and precon- centration of the cited elements from rocks. The analytical strategy which was actually used was presented in consider- able detail and matrix effects through all stages of the preparation methods were assessed and their effect on accuracy investigated. Detection limits by TXRF were in the range 1.2-2.0 ng ml-l for a 1000 s live time. Improvements in the calibration of XRF methods for geochemical analysis are constantly sought after. The Open University UK group (901C4098) have advocated the addition of pure silica to ore samples prior to fusion using a 1 + 1 + 12 sample to silica to flux ratio. Although the additional dilution causes a reduction of sensitivity this was not a problem in practice and the ability to use a routine silicate rock calibration was a major benefit.The same group have critically evaluated Zr data for silicate RMs and concluded that the available data were inadequate for the production of a reliable calibration even when Thiel's incomplete method was used to minimize the bias caused by high inaccurate concentration values. A standard additions method was used to determine more reliable values for some RMs and a reliable calibration could then be constructed. Eddy and Jacobs (911'259 1) have described the preparation of fusion bead standards for the generation of a single calibration which was suitable for the analysis of a great diversity of materials.The method and calibration was used for determination of analytes over the whole 0- 100% range in minerals ores and slags of Al Ca Cr Fe Mg Si Ti and V. The problems encountered in obtaining sufficient reliable standards for the determination of Nb and Zr in rocks has been circumvented by the use of SSMS for the certification of 16 international rock standards (9 1/295). Careful attention to reducing counting error and spectrum interference and to correction of matrix effects led to these workers obtaining an accurate calibration based on 13 standards. Precision for the determination of Nb was better than 0.4 pg g-l (95% confidence) and standard error was 0.9 and 2.7 pg g-l for Nb and Zr respectively.An XRF method for the determination of trace elements in iron rich samples has been reported (9 1/1555). The method used a Compton scatter correction procedure and was applicable over a concentration range from 0 to 70% Fe203. For the determination of Br C1 P and S in peat for which there are few standards available Anderson (90/3977) recom- mended a standard additions method. A multi-element standard additions was used and the linearity of the calibration graphs was validated by calculations based upon the Sherman equations. Once the necessary matrix correc- tions were applied the results showed good agreement with those from other instrumental methods. The use of a simultaneous WDXRF spectrometer for the direct multi-element analysis of UK coals has been described (91/2233).Of particular interest to these workers were the results for C1 P and S which were used to calculate combustion characteristics and price of the fuel and the ability of the equipment to provide up to 5000 determina- tions per day. Methods for calibration and sample prepara- tion had to take account of variations in mineralogy and particle size and detection limits were 0.05 0.01 and 0.002% m/m for major (A1 and Si) minor and trace (P) elements respectively. The exploitation of EDXRF spectrometry for the determi- nation of trace elements in geological samples has been reviewed by Potts et al. (90/2 198). The characteristics of EDXRF and WDXRF were compared and it was shown that although resolution and peak separation at energies below 5 keV were much better for WDXRF systems peak deconvolution software was able to correct the peak overlaps and at energies >15 keV EDXRF systems had better resolution.The optimization of EDXRF excitation conditions and use at energies >7 keV provided detection limits which were shown to be comparable to those from reported values for WDXRF systems. When a cobalt anode X-ray tube and iron beam filter were used the detection limits for Ba Cr and V (31 7 and 5 pg-l respectively) by EDXRF were at least as good as by WDXRF. An EDXRF system fitted with a gadolinium secondary target has been used to optimize the determination of Sn using the K X-ray emission line (90/327 1). Sample preparation was simple and the Compton Gd K a backscatter peak from the gadolinium target was used for simultaneous correction of variation in packing density and matrix effects.The detection limit for Sn was 2 ppm (500 s live time) which was four times worse than for GFAAS but overall the EDXRF method was preferred for its simplicity precision and wide calibration range. In their report of the use of an EDXRF spectrometer equipped with multiple secondary targets Roca and Bayon (9112516) found it necessary to use four different secondary targets and a fifth acquisition with direct excitation for the determination of 14 elements in each silicate rock sample. 2.4.4. In dust rial materials Warren (9014 164) has reviewed the role of XRF spectrome- try in the plastics and petrochemicals industry and noted the importance of advances in qualitative analysis and FP correction software.In addition increasing use of XRF equipment in the works environment and on-line was identified. One example where XRF spectrometry was particularly valuable was the direct analysis of polymer granules using the intensity of the Compton backscatter peak to correct for differences in packing density. In their short review of the application of atomic spectrometry to the analysis of advanced materials Marshall and Franks (91/837) concluded that XRF was well suited to the bulk elemental analysis of solid materials but was limited to concentrations >1 ppm and by the availability of solid standards. They indicated that laser ablation ICP-MS was likely to become increasingly important for solids analysis as a consequence of its superior sensitivity and that although TXRF was a promising technique for surface and interfacial analysis TXRF may be fatally flawed by its requirement for a flat analytical surface.The most recent Atomic Spectrometry Update on industrial materials was published during the present review period (9 1/27 19). A combination of TXRF ICP-MS and d.c. arc OES measurement has been recommended for quality control of trace elements in silica used in the electronics industry (91/111). The ICP-MS method had the highest sensitivity but required careful sample preparation whilst TXRF was particularly applicable to non-destructive analysis of sur- face contamination on solids. Klockenkaemper et al. (9 1 / 1 64) have used XRF to determine Co and Cu implanta- tion doses in the 1 pm surface layer of silicon wafers.Detection limits were 3 x 1014 atoms cm-* and calibration was accomplished using standards prepared by spreading spiked gelatine on cleaned uncoated wafers. La1 et al. (91/2588) have used EDXRF and PIXE methods for the analysis of high-temperature superconduct- ing materials. Thin films of YBa,Cu30 were deposited on various substrates such as thin Mylar sheet and thick substrates of magnesia and alumina. On the thin substrates the detection limits by EDXRF were 1000,3000 and 600 ngJOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY OCTOBER 199 1 VOL. 6 253R for Ba Cu and Y respectively and the corresponding limits for PIXE were 800 20 and 70 ng. Similar types of thin film material have been analysed successfully by EDXRF using either an FT calculation based upon one similar standard or on a suite of standards prepared by depositing known amounts of the analyte elements on filter discs (91/1532).Major elements in rutile and ilmenite were fused and then digested before being diluted to volume with an aqueous solution containing Cr as the internal standard (9013301). The resulting liquid was analysed by TXRF using a calibration graph which was linear only up to 1.7 ppm of Fe and 5 ppm of Ti. The complexity of the digestion coupled with the TXRF method seems rather inappropriate for analytes in materials which can readily and accurately be determined in the original solid by conventional XRF procedures. A simple EDXRF method has been developed for the determination of V in petroleum coke (91/333). The coke was ground mixed 1+1 with corn starch then a 0.5 g portion was pressed to form a pellet of about 0.1 g cm-2 mass thickness.A single calibration standard was used and absorption corrections were carried out by a simplified FP calculation method. Results showed close agreement with those obtained using GFAAS and although precision using the in-house radioisotope-excited EDXRF instrument was 3-5 times worse than by GFAAS the speed and simplicity of the EDXRF method and its ability to determine additional elements simultaneously made it an attractive alternative to AAS. The use of WDXRF spectrometry for the analysis of special materials used in the nuclear industry has been described by Hasany et al. (90/3432). They determined many elements in steels copper alloys zircalloy zirconium oxide and zircon ore and found good agreement with results from AAS ICP-OES and NAA.In most instances accuracy and precision were better than 4 and 2% respectively and the detection limit for Hf in a zirconium matrix was 40 pg g-l. Hoffman et al. (9013431) have investigated the determination of U in various matrices by EDXRF spectro- metry using several different radioisotope excitation sources. Detection limits for U were 20 pg cm-2 on filter- paper when using a 57C0 source and 10-100 pg g-' when using other sources and bulk samples. Martell and Hansel (9 1/454) developed a method based on anion-exchange separation and XRF spectrometry for the determination of Ga in plutonium. The RSD over the range 0.2-1% Ga was 0.36%. The non-destructive nature of XRF spectrometry makes it particularly attractive for the analysis of antiquities and archaeological relics. Mirti (9 1 / 1 1 34) included XRF in a discussion of the analytical techniques used in art and archaeology.An absolute standardless quantitative method has been proposed for the analysis of copper alloys (90/3272). The results were compared with those obtained by PIXE and detection limits using the XRF method are not encouraging (e.g. 1.26% Fe 0.37% Ni and 0.17% Pb in a bronze) which could be a consequence of instrument limitations and/or difficulties encountered in spectrum processing. In their SRXRF analysis of ancient Gaullish coins Brissaud et al. (91/356) found surface enrichment of Ag to be a problem as is the case with ageing of other ancient coinage metals.This serves as a useful reminder that no matter how complex and expensive the XRF instrumentation may be the form in which the sample is presented to it frequently constitutes the major analytical limitation. 2.4.5. Environmental analysis Maenhaut (91/1068) included XRF methods in an exten- sive review of analytical techniques for the determination of trace elements in the atmosphere. The Atomic Spectrome- try Update on environmental analysis (901349 1) included a section on developments in X-ray techniques. Kumar et al. (9111031 91/1274) have used radioisotope- excitation EDXRF spectrometry for the determination of 16 elements in air particulates. Detection limits were in the range 0.7-331 mg m-3 and the elemental data were used in a cluster analysis in order to identify specific sources of pollution. The same equipment was also used for the determination of Ca Fe Sr and Zn in ground water for which detection limits were around 0.1 pg ml-l after a preconcentration stage.Commercially available radioiso- tope-excited EDXRF instrumentation was used by An- selmo and Ehman (90/3275) for the determination of Pb in airborne particulate material collected on 37 mm glass-fibre filters. The Pb Lp line was used and the calibration was found to be linear only up to 5 pg cm-2 but that a quadratic fit could successfully be used to extend the range of 42 pg cm-2 of Pb. For the same type of analysis Galloo et al. (9013949) used almost 700 samples analysed by AAS to establish the Pb calibration and reported a detection limit of 0.02 pg m-3 of Pb in air.Thomson (9 1/456) has reported the interesting technique of direct XRF analysis of air to determine the concentration of heavy metals in aerosols. Such aerosols were reported as being widely found in battlefield conditions as by-products or active ingredients of pyrotechnics explosives anti- personnel munitions and rocket fuels. Instead of collecting the aerosol and analysing the static sample the X-ray beam was used to excite XRF in the ambient air mass and elements with 2>24 were directly determined in the air. Initial tests showed that aerosol concentrations from a few mg to several g m-3 could successfully be determined in less than 1 min. It was indicated that sensitivity could readily be improved if higher X-ray doses could be tolerated in the region of the analysis.With careful radiation shielding such a technique could find use in the real-time monitoring of heavy metals in the atmosphere of some workplace locations. 2.4.6. Biological materials The suitability of TXRF for multi-element determina- tions in human serum and whole blood has been investi- gated. Prange et al. (90/3422) used microwave digestion in nitric acid for sample preparation in a simple TXRF method for the determination of 11 elements in samples from a number of patients. Detection limits were in the range 0.08-0.2 pg rnl-l accuracy was in the range 2-15% and the RSD was better than 1Ooh in all cases. When the iron and salt in the whole blood matrix were removed the detection limits were improved to 2 ng ml-l and Mn and Ni could then also be accurately determined.Knoechel's group (9 1/ 1505) used low-temperature plasma ashing in an oxygen atmosphere to decompose blood samples. The residues were digested and the matrix then removed by ion exchange prior to TXRF determination of 16 ele- ments using both Mo and W excitation. Detection limits were in the range 0.6-2 ng ml-l after the matrix removal procedure compared with for example 36 ng ml-l for Cu without matrix removal. The in vivo measurement of bone Pb using K line XRF spectrometry has been further investigated by the Birmingham UK group. Improvements in measurement precision have been made by changing the collimation to reduce spectrum background and by incorporating pile-up rejection circuitry in the EDXRF measurement system (9111385).Once pile-up correction was made the back- ground under the Pb L& peak was reduced by a factor of five allowing the Pb Lj31,2 peak to be included with the Ka254R JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY OCTOBER 199 1 VOL. 6 peaks thereby improving counting precision. In a collabo- rative study with a Swedish group two XRF methods and three different bone sites have been evaluated for their ability to provide data for the study of chronic Pb exposure (91/1189). The methods used were 57C0 excitation and finger analysis or lo9Cd excitation for both tibia and calcaneus analysis. All methods showed close agreement of results and shared the ability to identify chronic occupa- tional Pb exposure in subjects. The lo9Cd measurement methods were the most precise and all methods gave similar radiation skin doses although the lo9Cd methods gave highest whole body dose.The whole body dose however was still small and likely to be reduced further with instrument improvements. A thin-film EDXRF technique has been used for the quantitative determination of Cd Cu Fe Ni Se and Zn in leukaemic single-cell populations (9 1/2628). As few as 400 000 cells could be used to provide valuable quantitative data. Evans et al. (9 1 / 1040) have compared XRF near-IR and glucose-release methods for the rapid determination of glucosinolate in oilseed rape. For the XRF method 20 g of seeds were milled in a coffee blender then gently pressed into an aluminium sample cup for measurement of the S Ka line in a sequential WDXRF instrument. Calibration was achieved by means of five measured samples and the samples were then analysed in less than 5 min each.Although the XRF method did not directly measure glucosinolate it required fewer standards than the near-IR method and was more rapid than the glucose-release method and was likely to remain the method of choice. The rapid determination of S in wheat grain by WDXRF has been reported by Griffiths et al. (91/1166). The whole grain was ground in a laboratory hammer mill and 1 g of this flour was backed with 3 g of cellulose powder and pressed to form a robust pellet. Further grinding to reduce particle size was not found to be necessary and cellulose backing was preferred to using boric acid as the former more closely matched the flour matrix.The total analysis time was less than 5 min and method precision was around 2% at 0.13% S. The Atomic Spectrometry Update on Clinical and Bio- logical Materials Foods and Beverages (9013500) included a section on developments in XRF. 2.4.7. Chemical effects Gohshi and co-workers (9112207) have employed a two- crystal WDXRF spectrometer to measure the Ka and Kp spectra for Sc from scandium metal and scandium com- pounds. These workers were able to assign a number of chemical shifts and showed that the Sc Ka chemical shift was strongly correlated with the electronegativity of the neighbouring atoms. LaBrecque and Rosales (90/4157) used an in-house radioisotope EDXRF system to measure Co KpIKa ratios for a number of calcination products of cobalt carbonate.They reported differences of up to 10% in the absorption- corrected KpIKa ratio and suggested these to be the result of differences in molecular structure. More surprising were the reported shifts in energy of up to 50 eV observed for the Co yB line. These shifts were attributed to chemical effects or instrument limitations although the latter seems the more plausible explanation. A WDXRF spectrometer has been used for the determi- nation of the various aluminium phases in steel (90/3273). The A1 K/3 lines from aluminium aluminium nitride and alumina were measured but mutual overlap of the lines was such that the detection limit for each phase was 0.06% even when peak deconvolution was used. This detection limit was too high for modern steel analysis and the analysis time of 3 min was too long for process control which led to the conclusion that XRF was unlikely to supplant conventional phase analysis.A WDXRF spectrometer using a low-energy electron- excitation source has been used for the non-destructive chemical state analysis of thick films and surface layers (91/2587). The probed depth was controlled by the energy of the incident electrons in the range 3-12 keV and layers 10-1000 nm thick were analysed. The large shifts in the low-energy characteristic lines were used to determine the chemical state of elements such as A1 in alumina layers on aluminium metal Cr and Fe in oxide layers on steel and B in silica layers. LOCATION OF REFERENCES The full list of references cited in this Update have been published as follows 9012278-90/3591 J.Anal. At. Spectrom. 1990 5(7) 278R-321R. 90/3 592-90/4 179 J. Anal. At. Spectrom. 1990 5( 8) 36 1 R-378R. 91/1-91/825 J. Anal. At. Spectrom. 1991 6(1) 41R-68R. 91/826-91/C1687 J. Anal. At. Spectrom. 1991 6(3) 109R-136R. 91/C1688-91/2702 J. Anal. At. Spectrom. 1991 6(4) 153R-185R. 91/2703-91/C2929 J. Anal. At. Spectrom. 1991 6(5) 221R-227R. Abbreviated forms of the literature references quoted (excluding those to Conference Proceedings) are given on the following pages for the convenience of the readers. The full references names and addresses of the authors and details of the Conference presentations can be found in the appropriate issues of JAAS cited above.JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY OCTOBER 1991 VOL. 6 Abbreviated List of References Cited in Update 255R 9012337. Fresenius 2.Anal. Chem. 1989 333 561. 9012375. Microbeam Anal. 1989 24 3 1 1. 9012377. Micro- beam Anal. 1989 24 319. 9012384. Microbeam Anal. 1989 24 591. 9012719. J. Phys. E Sci. Instrum. 1989 22 154. 9013269. X-Ray Spectrom. 1989,18 15 1. 90/3270. X- Ray Spectrom. 1989 18 259. 9013271. X-Ray Spectrom. 1989 18 139. 90J3272. X-Ray Spectrom. 1989 18 157. 9013273. X-Ray Spectrom. 1989 18 177. 9013274. X-Ray Spectrom. 1989 18 143. 90J3275. Spectroscopy (Eugene Oreg.) 1989 4(6) 41. 9013286. At. Spectrosc. 1989 10(4) 97.9013289. At. Spectrosc. 1989 10(4) I 12.9013301. Appl. Spectrosc. 1989 43 702. 9013309. Appl. Spectrosc. 1989 43 962. 9013364. Bunseki Kagaku 1989 38 T123. 9013422. Fresenius 2. Anal. Chem.1989 335 914. 9013430. J. Radioanal. Nucl. Chem. 1989 131 399. 9013431. J. Radioanal. Nucl. Chem. 1989 132 121. 9013434. J. Radioanal. Nucl. Chem. 1989 136 353. 9013437. Adv. X-Ray Anal. 1988 31 439.9013438. J. Vac. Sci. Technol. A. 1989 7 3305. 9013471. Fenxi Shiyanshi 1989 8(4) 62.9013489. J. Anal. At. Spectrorn. 1990,5 75. 9013491. J. Anal. At. Spectrom. 1990 5 1R. 9013497. J. Anal. At. Spectrom. 1990 5 135. 9013500. J. Anal. At. Spectrom. 1990 5 75R. 9013533. Analusis 1989 17 437. 9013538. Rev. Sci. Instrum. 1989 60(7 Pt. 2B) 2456. 90/3539. Adv. X-Ray Anal. 1988,31,479. 9013540. Adv. X- Ray Anal. 1988,31,507.90/3557. Anal. Proc. 1990,27,8 1. 9013586. Talanta 1990 37 135. 9013590. Talanta 1990 37 365.9013949. Analusis 1989,17 576.9013977. Talanta 1990 37 185.9014003. Fresenius 2. Anal. Chem. 1989 335 755. 9014007. Fresenius 2. Anal. Chem. 1989 335 900. 9014113. Isotopenpraxis 1989,25 349.9014117. Int. J. Mass. Spectrom. Ion Processes 1989,94 10 1. 90141 19. Int. J. Mass Spectrom. Ion Processes 1989 93 303.9014157. J. Anal. At. Spectrom. 1990 5 269. 9014159. Analyst 1990 115 813. 9014164. Anal. Proc. 1990 27 186. 9014165. Bunseki Kagaku 1989 38 468. 9014174. Appl. Opt. 1990 29 477. 9014175. Appl. Opt. 1990 29 483. 9112. J. Anal. At. Spectrom. 1990 5 293. 9113. J. Anal. At. Spectrom. 1990 5 301. 91/8. J. Anal. At. Spectrorn. 1990 5 151R. 9119. Analyst 1990 115 9 1 1. 91122. Appl. Spectrosc. 1989 43 1283. 91173. Anal. Sci. 1989 5 435. 91174. Anal. Sci. 1990 6 183. 911111. Mikrochim. Acta 1989 3 413. 911114. Microbeam Anal.1989 24 186. 911142. Adv. X- Ray Anal. 1989 32 1. 911143. Adv. X-Ray Anal. 1989,32 39. 91/144. Adv. X-Ray Anal. 1989,32 45.911145. Adv. X- Ray Anal. 1989,32,49.91/146. Adv. X-Ray Anal. 1989,32 191. 911147. Adv. X-Ray Anal. 1989,32 205.911148. Adv. X-Ray Anal. 1989 32 211. 911149. Adv. X-Ray Anal. 1989 32 221. 911151. Adv. X-Ray Anal. 1989 32 233. 911152. Adv. X-Ray Anal 1989 32 239. 91/160. Anal. Chem. 1990 62 303R. 911164. Anal. Chem. 1990 62 1674. 911201. X-Ray Spectrom. 1989 18 253. 911202. X- Ray Spectrom. 1990,19 3.911203. X-Ray Spectrom. 1990 19 45. 911204. X-Ray Spectrom. 1990 12 53. 911205. X- Ray Spectrom. 1990 19 67. 911227. Fresenius Z. Anal. Chem. 1989 335 124. 911228. Fresenius J. Anal. Chem. 1990 336 226. 911242. Alzal. Chim. Acta 1990 228 333. 911294. Chem.Geol. 1989 78 1. 911295. Chem. Geol. 1990 81 1. 91/333. J. Radioanal. Nucl. Chem. 1990 144 327. 911334. J. Radioanal. Nucl. Chem. 1990 144 407. 911352. Nucl. Instrum. Methods Phys. Res. Sect. B 1990 45 527. 911356. Nucl. Instrum. Methods Phys. Rex Sect. B 1990,49 305.911426. Lihua Jianyan Huaxue Fence 1989 25 66. 911454. Report 1988 LA-11435; Order No. DE89004542 13 pp. 91/456. Report 1989 BRL-TR-3003; Order No. AD-A209077 47 pp. 911699. Geostand. Newsl. 1989 13 197. 911704. Geostand. Newsl. 1990 14 119. 911707. Geostand Newsl. 1990 14 197. 911719. Appl. Spectrosc. 1989 43 912. 911726. J. Geochem. Explor. 1990 37 1. 911727. Chem. Geol. 1990 83 1. 91/728. Chem. Geol. 1990 83 89. 911729. Chem. Geol. 1990 83 105. 911730. Chem. Geol. 1990 83 119. 91/731.Chem. Geol. 1990 83 133. 911826. Analyst 1990 115 1019. 911835. Analyst 1990 115 1329.911837. Anal. Proc. 1990 27 240. 911839. J. Anal. At. Spectrom. 1990 5 339. 911840. J. Anal. At. Spectrom. 1990,5 35 1.911850. J. Anal. At. Spectrom. 1990 5 41 9. 911852. J. Anal. At. Spectrom. 1990 5 425. 92/853. J. Anal. At. Spectrom. 1990 5 431. 911854. J. Anal. At. Spectrom. 1990,5,437.911855. J. Anal. At. Spectrom. 1990 5 445. 911856. J. Anal. At. Spectrom. 1990 5 451. 911857. J Anal. At. Spectrom. 1990 5 457. 911858. J. Anal. At. Spectrom. 1990 5 463. 9111004. Chem. Anal. (N. Y.) 1988 99 175. 9111032. Indian J. Technol. 1988 26 524. 9111040. J. Sci. Food Agric. 1989 49 297. 9111054. Kagaku Gijutsu Kenkyusho Hokoku 1989 84 643. 9111055. Kankyo Gijutsu 1989 28 564. 9111068.NATO ASI Ser. Ser. C 1989 268 (Control Fate Atmos. Trace Met.) 259. 9111075. Proc.-Electrochem. Soc. 1990 90-11 (Anal. Tech. Semicond. Mater. Process Char- act.) 355. 9111084. Radioisotopes 1989 38 15 1. 9111085. Radioisotopes 1989,38 153.9111133. CRC Crit. Rev. Anal. Chem. 1989 21 225. 9111134. Crit. Rev. Anal. Chem. 1988 20 221. 9111148. GIT Fachz. Lab. 1989 33 971. 9111159. Indian J. Technol. 1989,27 366. 91/1166. J. Sci. Food Agric. 1990 51 27. 9111189. Phys. Med. Biol. 1989 34 1833. 9111194. Proc. SPIE-Int. SOC. Opt. Eng. 1989 1186 (Surf. Interface Anal. Microelectron. Mater. Process. Growth) 22. 9111201. Rigaku Denki Janaru 1989 20(1) 28. 9111205. Stahl Eisen 1989 109 1161. 9111217. X-sen Bunseki no Shinpo 1988 20 16 1. 9111218. X-sen Bunseki no Shinpo 1988 20 193.9111223. ZKG Int. Ed. B 1989 42 294. 9111250. Chem. Anal. (N.Y.) 1989 105 288. 9111373. Kenkyu Hokoku-Asahi Garasu Kogyo Gijutsu Shoreikai 1988 53 65. 9111385. Phys. Med. Biol. 1989 34 1295. 9111391. Radioisotopes 1989 38 179. 9111404. Zap. Leningr. Gorn. Inst. im. G. V. Plekhanova 1987,111 109. 9111427. J. Anal. At. Spectrom. 1990 5 61 1.9111428. J. Anal. At. Spectrom. 1990 5 619. 9111429. J. Anal. At. Spectrom. 1990 5 623. 9111430. J. Anal. At. Spectrom. 1990 5 627. 9111436. J. Anal. At. Spectrom. 1990 5 243R. 9111485. Anal. Sci. 1989,5 355.9111502. Fresenius 2. And. Chem. 1989,334,723.9111503. Fresenius Z. Anal. Chem. 1989,335 680.9111505. Fresenius 2. Anal. Chem. 1989 335 855. 9111516. Fresenius J. Anal. Chem. 1990 337 55 1. 9111532. Appl. Spectrosc.1990 44 826. 9111536. Appl. Spectrosc. 1990,44,998.9111545. Anal. Chem. 1989 61 1851. 9111553. X-Ray Spectrom. 1990 19(3) 97. 9111555. X-Ray Spectrom. 1990 19(3) 117. 9111557. X- Ray Spectrorn. 1990 19(3) 145. 9111558. X-Ray Spec- trom. 1990 19(3) 159. 9111564. Nucl. Instrum. Methods Pys. Res. Sect. B 1990 49 11 5. 9111565. Nucl. Instrum. Methods Phys. Res. Sect. B 1990 49 518. 9111566. Nucl. Instrum. Methods Phys. Rex Sect. A 1990 292 141. 9111567. Nucl. Instrum. Methods Phys. Res. Sect. A 1990 292 147. 9111568. Guangpuxue Yu Guangpu Fenxi 1989 9(6) 40. 9111580. J. Radioanal. Nucl. Chem. 1990 145 259. 9111586. Adv. X-Ray Anal. 1989,32,69.91/1587. Adv. X-Ray Anal. 1989,32,77.91/1588. Adv. X-Ray Anal. 1989 32 105.9111589. Adv. X-Ray Anal. 1989,32 1 15.9111592. Adv. X-Ray Anal. 1989,32 185.9111609. Bunseki Kagaku 1990 39 T107. 9112153. Microbeam Anal. 1989 xvii. 9112154. Microbeam Anal 1989 17-22. 9112155. Micro- beam Anal. 1989 35-42.9112159. Microbeam Anal. 1989 26 1.91/2161. Microbeam Anal. 1989,269.9112162. Micro- beam Anal. 1989 273. 9112167. Microbeam Anal. 1989 293. 9112172. J. Appl. Phys. 1990 68 1421. 9112180.256R JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY OCTOBER 1991 VOL. 6 Microbeam Anal. 1989 353. 9112182. Microbeam Anal. 1989 359. 9112183. Microbeam Anal. 1989 364.9112185. Microbeam Anal. 1989 370. 9112186. Microbeam Anal. 1989 586. 9112193. Spectrochim. Acta Part B 1990,45,37. 9112198. Spectrochim. Acta Part B 1990,45,249.9112201. Spectrochim. Acta Part B 1990,45 333. 9112207. Spectro- chim. Acta Part B 1990 45 463. 9112208. Spectrochim. Acta Part B 1990 45 499. 9112211. Spectrochim. Acta Part B 1990 45 527. 9112214. Spectrochim. Acta Part B 1990,45 58 1.9112215. Spectrochim. Acta Part B 1990,45 6 13. 9112219. Talanta 1990 37 8 19.9112231 Anal. Proc. 1990 27 324. 9112233. Analyst 1990 115 1397.9112242. Zh. Anal. Khim 1989 44 2002. 9112243. Zh. Anal. Khim 1989 44 2157. 9112247. Zh. Anal. Khim. 1990 45 858. 9112248. Zh. Anal. Khim. 1990 45 1197. 9112251. Anal. Sci. 1990 6 157. 9112252. Anal. Sci. 1990 6 191. 9112254. Anal. Sci. 1990 6 397. 9112256. Anal. Chim. Acta 1989 224 83. 9112257. Anal. Chim. Acta 1990 228 1. 9112259. Anal. Chim. Acta 1990 229 157. 9112260. Fresenius 2. Anal. Chem. 1989,335 390. 9112261. Fresen- ius Z. Anal. Chem. 1989 335 478. 9112262. Fresenius Z. Anal. Chem. 1989 335 675. 9112263. Fresenius Z. Anal. Chem. 1989 335,945.9112264. Surf Interface Anal. 1989 393 400. 9112265. Surf. Interface Anal. 1989 14 491. 9112266. Surf. Interface Anal. 1989,14 505.9112267. Surf Interface Anal. 1989 14 537. 9112269. Surf Interface Anal. 1989 14 552. 9112270. Surf Interface Anal. 1989 14 555. 9112271. Surf Interface Anal. 1989 14 595. 9112272. Surf Interface Anal. 1989 14 598. 9112273. Surf Interface Anal. 1989 14 623. 9112274. Surf. Interface Anal. 1989 14 647. 9112276. Surf Interface Anal. 1989 14 725. 9112277. Surf Interface Anal. 1989 14 739. 9112278. Surf Interface Anal. 1989 14 75 1-91/2279. Surf Interface Anal. 1989 14 771. 9112280. Surf Interface Anal. 1989 14 781. 9112282. Surf Interface Anal. 1989 15 15. 9112284. Int. J. Mass Spectrom. Ion Processes 1989 91 229. 9112285. Int. J. Mass Spectrom. Ion Processes 1989 91 241. 9112286. Int. J. Mass Spectrom. Ion Pro- cesses 1989 94 261. 9112287. Int. J. Mass Spectrom. Ion Processes 1989 95 1. 9112289. Int. J. Mass Spectrom. Ion Processes 1990 96 RI. 9112290. Nucl. Instrum. Methods Phys. Res. Sect. A 1989 281 547. 9112291. Nucl. Instrum. Methods Phys. Res. Sect. B. 1989 43 507. 9112292. Nucl. Instrum. Methods Phys. Res. Sect. B 1990 45 565. 9112293. Nucl. Instrum. Methods Phys. Res. Sect. B 1990 45 570. 9112294. Nucl. Instrum. Methods Phys. Res. Sect. B 1990 45 575. 91/2295. Nucl. Instrum. Methods Phys. Res. Sect. B 1990 45 592. 9112301. J. Vac. Sci. Technol. 1989 48 3065. 9112302. J. Vac. Sci. Technol. 1989 A7 3 126. 91/2303. J. Vac. Sci. Technol. 1990 A8,93.91/2306. Conf. Ser. Inst. Phys. 1989 No. 100 Sect. 7 519. 9112314. Pediatr. Res. 1988 23 495. 9112315. Pediatr. Res. 1988 24 20.9112316. Pediatr. Res. 1989,26,250.9112317. Earth Planet. Sci. Lett. 1989 96 147. 9112318. Earth Planet. Sci. Lett. 1989 96 134. 9112319. NIST Spec. Publ. 1989 775 (Laser Induced Damage Opt. Mater) 183. 9112320. Thin Solid Films 1989 174 1 17. 9112322. Mem. Sci. Technol. 1990 1 255. 9112323. Second Int. Conf Nat. Glasses Prague 1987 Ed. Konta J. Publ. Charles Univ. Prague 1989 391. 9112326. Jpn. J. Appl. Phys. 1990 29 386. 9112329. Anal. Chem. 1990 62 972. 9112330. Radiochim. Acta 1989 48 165.9112331. Mater Chem. Phys. 1990,24 327. 9112332. J. Anal. Chem. USSR Engl. Transl. 1989 44 561. 9112335. Microbeam Anal. 1989,24,264.9112336. Phys. Rev. A 1990 41 2566. 9112338. Nucl. Instrum. Methods Phys. Res. 1990 B45 561. 9112339. Rev. Sci. Instrum. 1989 60 3560. 9112343. Biophys. J. 1989 56 437. 9112344. John Wiley. 9112345. Rep. Invest. Bur. Mines 1989 RI 9249. 9112350. J. Appl. Phys. 1989 66 6055. 9112351. Quat. Res. 1989 31 377. 9112352. Globigerina bulloides Quat. Res. 1989 31 381. 9112354. J. Opt. SOC. Am. B Opt. Phys. 1990 7 3. 9112357. Appl. Spectrosc. 1989 43 1164. 9112358. Pract. Spectrosc. 1990 8 413. 9112359. Int. J. Environ. Anal. Chem. 1990 38 427. 9112361. Certif. Ref Mater. Proc. ISCW89 1989 292. 9112363. Science in China (Series B) 1989 32 1409. 9112365. J. Phys. Condens. Matter I 1989 SB99.9112366. Vacuum 1989 39 1077. 9112367. J. Mater Res. 1990 5 578. 9112378. Fenxi Shiyanshi 1990 9(2) 22. 9112394. X- Ray Spectrom. 1990 19 177. 9112395. X-Ray Spectrom. 1990 19 185. 9112396. X-Ray Spectrom. 1990 19 203. 9112398. X-sen Bunseki no Shinpo 1989,21 123.9112400. Spectroscopy (Eugene Oreg.) 1990 5(6) 26. 9112414. Energy Sources 1990 12 25 1.9112437. SIA Surf Interface Anal. 1990,16,429.9112455. Anal. Chem. 1990,62 101R. 9112481. Prog. Anal. Spectrosc. 1989 12 433. 9112503. Fresenius J. Anal. Chem. 1990,336,99.91/2512. Fresenius J. Anal. Chem. 1990 337 577. 9112516. An. Quim. 1990 86 230. 9112538. He Huaxue Yu Fangshe Huaxue 1989 11 118. 9112547. Int. Steel Met. Mag. 1990 28 307. 9112583. Nippon Kinzoku Gakkai-shi 1990 54 575. 9112587. Phys. Scr. 1990 41 878. 9112588. Pramana 1990 34 L377. 9112591. Rep. MINTEK 1990 M402. 9112592. Rev. Sci. Instrum. 1990 61 2121. 9112595. SINTEF Rep. 1989 STF36 A89026. 9112614. Adv. X-Ray Anal. 1989,32 197. 9112628. Biol. TraceElem. Res. 1989 23 65. 9112648. Hyomen Kagaku 1990,11 189. 9112650. Isotopenpraxis 1990 26 34 1. 9112681. Shimadzu Hyoron 1990,47 15 1.9112716. J. Anal. At. Spectrom. 1990,5697. 9112719. J. Anal. At. Spectrom. 1990 5 323R.